GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter X Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted
Musings of an Energy Nerd

Installing Rigid Foam Above Roof Sheathing

An updated look at roof assemblies that include a continuous layer of exterior insulation

Insulation performs best when it is installing in a continuous layer. If this continuous insulation is on the exterior side of the sheathing, it helps keep the sheathing warm and dry during the winter. [Image credit: Fine Homebuilding]

In a traditional house with a gable roof, attics were usually vented and unconditioned, with insulation installed on the attic floor.

These days, however, builders often choose to install insulation that follows the sloping roofline. The reasons for this trend are many:

  • This approach can be used to create an unvented conditioned attic, allowing attic ducts to be installed inside a home’s thermal envelope.
  • Many modern designs include cathedral ceilings, and cathedral ceilings need insulation that follows the roof slope.
  • In a Cape Code house, or any house with a half story on the top floor, insulation should usually follow the roof slope.

There are many ways to insulate a sloped roof. The roof assembly can be vented or unvented; insulation can be installed above the roof sheathing, between the rafters, under the rafters, or in a combination of these different locations. For a complete description of all the different ways to insulate a sloped roof, see “How to Build an Insulated Cathedral Ceiling.”

In 2015, I wrote an article (“How to Install Rigid Foam On Top of Roof Sheathing”) that explained how to create an unvented insulated roof with a continuous layer of rigid foam on the exterior side of the roof sheathing. It’s time to update the article with new advice.

An excellent way to insulate a roof

Why is installing exterior rigid foam such a good way to insulate a sloped roof? For at least two reasons: the rigid foam interrupts thermal bridging through the rafters, and it keeps the sheathing warmer and drier than it would be if all the insulation were on the interior side of the roof sheathing.

In recent years, building codes have increased minimum R-value requirements for ceilings and roofs. Code requirements vary by climate and jurisdiction, but in general, the…

GBA Prime

This article is only available to GBA Prime Members

Sign up for a free trial and get instant access to this article as well as GBA’s complete library of premium articles and construction details.

Start Free Trial


  1. maine_tyler | | #1

    "After taping the panel seams, install the roofing underlayment of your choice (for example, asphalt felt)."

    I know there has been some debate on the location of a wall wrb with exterior foam, but I wonder if the underlayment makes more sense above the foam, or above the second layer of sheathing if using that.
    Assuming it's main purpose is shedding bulk water, what possible advantage would it have below all that foam? It seems there are disadvantages to that location, such as reduced drying potential, not protecting the top sheathing layer, and allowing water to seep between foam seems.

    1. GBA Editor
      Martin Holladay | | #2

      Thanks for your comment. You're right.

      (Due to a cut-and-paste error during the writing process, my advice on where to install the roofing underlayment got garbled. I've corrected my error.)

  2. mikesmcp | | #3

    Thank you Martin for the updated article. I have a project in southern VT I have three questions about.
    Local, green, rough sawn, 1" board sheathing on a hip roof. Some undetermined type of air barrier immediately above the board sheathing. 4" reclaimed polyiso, then 2" EPS, staggered and top seams taped. 7/16 OSB fastened with long screws into rafters, tar paper, shingles. Under board sheathing, fiberglass insulation and rough sawn ship lap instead of sheet rock in between partly exposed rafters.

    1). I have a roll of black 6 mil plastic that has gone unused and taking up space in my shop for some time. Could I use this above the board sheathing instead of the other membranes you mentioned? The rafters are 3" thick, so I am hoping with some careful planning, I wont have too many missed screws leaving holes in the plastic.

    2). And since the underside of the sheathing will dry inwards, is my far from air tight ceiling of shiplap boards a solid approach? Shiplap boards will keep fiberglass insulation tight up against board sheathing.

    3). I was thinking of using kraft-faced only to help keep dust and fiberglass from falling through the cracks in the shiplap. I am thinking I do not need a vapor retarder there, so is there any other better ideas to accomplish the ship lap look without worry of dust falling through the cracks?

    1. GBA Editor
      Martin Holladay | | #5

      If I understand your question correctly, you wonder whether an old roll of black polyethylene that has been leaning against the wall in your shop for a few years can be used as "some undetermined type of air barrier" above green rough-sawn board roof sheathing. I'd say it's possible but risky.

      I've built with rough-sawn green lumber. The stuff twists and shrinks as it dries. It can cup and it can check. You'll have far more movement over the next few years than you would with OSB or plywood.

      Your polyethylene is 6 mils thick. It's easy to put holes in it, especially when you're working on a roof and walking on it as you install it. It makes a lousy air barrier.

      I know that your taped EPS is an air barrier, but it may not be perfect. (EPS can be hard to tape unless you invest in the very best types of European tape -- and you don't strike me as a "nothing but the best for me" type of shopper).

      With this type of roof assembly, nothing is more important than the air barrier. Don't skimp on this detail, or you'll live to regret it. You could use taped Zip sheathing as your air barrier, but that would cost money. If I were you, that's what I would do -- or I'd buy a self-adhered membrane to cover the rough boards.

      If you can manage to establish a robust air barrier above your roof sheathing, the fact that your finish ceiling is leaky doesn't matter. The kraft facing you propose is unnecessary but harmless.

      1. mikesmcp | | #7

        " you don't strike me as a "nothing but the best for me" type of shopper)."

        I got a good chuckle out of your observation. I can see why you got that impression, and there is some truth to that is some cases. For this case though, I really wont take any shortcuts that undermine the robustness of a roof system. Especially a roof like this with extra risks, compared to my preferred scenario of vented, unconditioned attic with loads of cellulose in the ceiling and walls.

        I have been reading your articles for a long time now. I had what I thought was a solid understanding of how to build house the "right way" in the 1980's. It has been a wild, frustrating ride since the day I decided to start using a lot more insulation, and found one of your articles during my research.

        I try to use rough sawn lumber every time I can because I mill it myself and there are less miles driven for transport than anything I could buy from any vendor. And I like to re use and re purpose anything I can (e.g. my black plastic I ended up with extra).

        After thinking this thru, I will probably use zip/taped, because I think it probably is the least risky and most robust.

        Question: If I did use dried boards for sheathing, I am concerned about using a "self-adhered membrane to cover the rough boards". I talked with a rep from Grace about the Ice and Water Shield years ago and he said they definitely do not recommend using that over any board sheathing due to a reaction which causes the membrane to turn to goo and drip through the cracks. He said only apply it to engineered sheet materials. Have you heard about this, and if so which self adhered membrane would you use instead?

        1. GBA Editor
          Martin Holladay | | #8

          The rep from Grace certainly knows more about Ice and Water Shield than I do. So your plan to install taped Zip sheathing -- my first choice -- is certainly the way to go. Good luck.

  3. pablo9 | | #4

    Much appreciate the updated article. We live in the Bay Area with a Mediterranean climate. We have a 1961 home with cathedral ceilings and a 2.5 in 12 roof pitch. There is currently some kind of built-up roof surface with minimal insulation above the 2" x 6" tongue and groove roof sheathing and none below it. There is a small, unvented, conditioned attic, but the ceiling is mostly exposed tongue and groove when looking up from the inside of the house.

    We need a new roof and would like to add a 2 1/2 to 3 inch polyiso nail base over the sheathing, such as Hunter Panels H-Shied NB. I was originally thinking of using a vented nail base, but that seems unwarranted given the mild climate and shallow roof slope. Would you concur with that approach?

    1. GBA Editor
      Martin Holladay | | #6

      Your approach will work. My only comment is that your R-value will be low. If you go with 2.5 inches of polyiso, you'll have an R-value of R-16. That isn't much. Consider installing a continuous layer of rigid foam under the nailbase to get more R-value.

      1. pablo9 | | #33

        Thanks, Martin. After your comment, I am now thinking about 3.5 inches of polyiso covered by 1/2 inch ply. I am also considering GAF Energy solar shingles. My understanding is that they can get quite hot on sunny days. Do you think doing so would create any problems on this type of unvented roof? The roofing firm is offering GAF's Golden Assurance warranty that includes 25 years of workmanship and materials with no proration.

        1. MartinHolladay | | #34

          In general, the fact that roofing may be somewhat hotter on an unvented roof assembly than on a vented roof assembly isn't particularly significant. (Other factors, most importantly the color of the roofing, have a greater effect on roofing temperature than the presence or absence of a ventilation channel under the sheathing.)

          So as long as GAF, the manufacturer of the solar shingle you are considering, has no problem with your proposed installation, I wouldn't worry.

  4. jimgove30 | | #9

    Hi Martin,
    Love your articles, and have been pouring over them the last year or so. This one's particularly timely, as we're about to build in the White Mountains, NH (Zone 6) and I'm now very concerned with my roof assembly detail (see attached). There is an area at the L in the roof that appears cannot be vented, as the detail states. I having a bit of trouble getting my team to answer my questions regarding venting and insulation requirements and details. If the roofline intersecting area cannot be vented, I assume I'm asking for trouble? Is the unvented, exterior insulation shown here about the only safe approach for us up in snowy, cold NH? Are the details to include interior ff polyiso adequate?

    1. paulrey27 | | #17

      Look up "valley vent" by DCI Products - I've not used this but does this achieve what you're after?

  5. GBA Editor
    Martin Holladay | | #10

    You're correct that the roof valleys in you design make venting very difficult. For a detailed discussion of this issue, see this article: "All About Roof Venting." (Scroll down to the section with this headline: "Tricky roofs to vent: roofs with hips or valleys and mansard roofs.")

    So, unless you want to adopt the "checkerboard approach" to venting (as described in the article I linked to), your options are either (a) closed-cell spray foam under the roof sheathing -- an unvented approach, or (b) a combination of rigid foam above the roof sheathing and some other type of insulation directly under the roof sheathing, as described in the article on this page (also an unvented approach).

    1. jimgove30 | | #28

      Martin, thanks for the response. Could the detail I attached work as an UNVENTED assembly? As I look at the detail shown in my plan, I wonder whether I misunderstood and the intent was to actually NOT ventilate the rafter bays, and only ventilate the underside of the metal roof. Is a ventilated underlayment enough space to allow any moisture to dry outwards through the sheathing and #30 felt? I've not seen this approach recommended anywhere else as a safe unvented option, but if it could dry to the exterior, could this work as detailed? If not, I'll pursue one of the options you've mentioned. Thanks again.

      1. MartinHolladay | | #31

        Q. "Could the detail I attached work as an UNVENTED assembly?"

        A. Yes. That's why I wrote (in my last answer to you), "Unless you want to adopt the 'checkerboard approach' to venting (as described in the article I linked to), your options are either (a) closed-cell spray foam under the roof sheathing -- an unvented approach, or (b) a combination of rigid foam above the roof sheathing and some other type of insulation directly under the roof sheathing, as described in the article on this page (also an unvented approach)."

        Q. "Is a ventilated underlayment enough space to allow any moisture to dry outwards through the sheathing and #30 felt?"

        A. There is nothing wrong with using vented underlayment above an unvented insulated roof assembly. But you can't expect this type of underlayment to allow damp sheathing to dry outward. (Outward drying is prevented in your case by your roofing choice -- standing-seam metal roofing. If you want outward drying, you would need to install vapor-permeable roofing like cedar shingles, concrete roofing tiles, or slate.)

        But an unvented roof assembly doesn't depend on outward drying to stay safe. Instead, the idea is to prevent the roof sheathing from getting wet in the first place.

        1. pardi001 | | #35

          Let me make sure I am not overlooking anything. Planning an unvented attic and would use 2” of closed cell followed by an additional layer of open cell to achieve adequate R value at acceptable cost. So this should prevent humidity in the attic from reaching the underside of the sheathing and especially to avoid ridge rot. Now, as you have pointed out, if the outside of the sheathing becomes wet, it would have to dry to the outside. And it’s best not to allow it to get wet in the first place. Now I am planning to have a metal roof and I was thinking about using 2x4s to create ventilation channels. I would use a WRB such as Mento over the sheathing which as high vapor permeability so that, if in spite of best efforts, the sheathing got wet, it could dry to the outside. Now if I add exterior roof insulation, it too would have to be vapor permeable. So mineral wool boards would work here if I can find someone to do it. Two questions: Some GPS board products claim to be somewhat permeable. Would they be a good choice? And focusing again on preventing the sheathing from getting wet in the first place, when I would secure 2x4s to the rafters, would I need to use a double-sided self annealing tape under the 2x4s to prevent breaching the air and water sealing provided by the Mento? Thanks!


          1. MartinHolladay | | #49

            Your plan for a vented assembly will work. However, it makes no sense to add additional insulation (in your case, the planned semi-rigid mineral wool) on the exterior side of a roof assembly, since all that moving air in the ventilation channel is exterior (cold-in-winter) air.

            As I explained in my article, "How to Build an Insulated Cathedral Ceiling," "Remember: if you choose to install rigid foam [or semi-rigid mineral wool] on top of your roof sheathing, don’t install ventilation channels under the roof sheathing; these two practices are incompatible."

        2. jimgove30 | | #48

          Hi Martin,

          Just to follow up on the ceiling approach for my project - As I've been trying to push the builder/architect to consider/design this assembly in a safe manner, they have responded with an unexpected and possibly more confusing recommendation of using parallel chord trusses for the entire roof, and are suggesting the option to insulate will essentially be closed cell spray foam. I think many of the assembly variations can still work, including rigid exterior insulation hybrids, and a vented assembly. Is there anything particularly unique or worrisome about this type of truss design assembly for efficiency? (Its about a 28" deep truss cavity) Does this alleviate any issues in the intersecting roofline venting? Or are there any benefits/drawbacks to this approach? Hopefully we'll get to a solution soon

          1. MartinHolladay | | #50

            Deep rafters are good, but you still have a few hurdles ahead of you.

            Remember that deep rafters alone won't solve the problem of venting a roof with valleys or hips -- you still end up with insufficient air flow in all of those valley rafters and hip rafters. Moreover, you have to worry about insulation slumping if you are attempting a vented assembly. (For more on insulation slumping, see "Insulating Scissors Trusses.")

            However, the rafters you plan on using are compatible with closed-cell spray foam used to create an unvented assembly.

  6. alanferg | | #11

    Martin, this is probably the #1 article I read every time I have a question or need a refresher on roof assembly. Thank you for sharing your knowledge and updating it.

  7. stevehallarchitecture | | #12

    It should be noted that IBC Climate Zones have been significantly updated across most of the US since the linked map. This affects required insulation values for a substantial number of counties, according to each state's adoption date and modification practices.

    1. GBA Editor
      Martin Holladay | | #14

      You're right. But note that few builders (if any) are working in jurisdictions where the 2021 IECC has been adopted. For those who are interested in more details on the changes to the climate zone map, I recommend this article: "Updated Climate Zone Map Reflects Warming Trend."

      More information on the 2021 IECC can be found here: "Understanding the Latest Energy Code."

      An online version of the new climate zone map -- one that is both high-resolution and easy to read -- has not yet become available, to the best of my knowledge. Here is a link to the best current version that I can find: New Climate Zone Map from 2021 IECC.

      1. stevehallarchitecture | | #16

        Martin, thanks for the article link. I guess I'm overly-sensitive to this because I'm in a region that is changing. As a designer, we always worry that projects starting in one code cycle might not permit until the next. (Frequently design pauses or holds for funding.)

        Ironically, our region is moving from CZ4 to 3, so it gets easier. And NC has rolled back IBC Energy Code requirements via modifications, so CZ4 has only ever needed R-38 (R-30 with insulated top plates) roof–ceiling. But supposedly NC roll-backs are going to end, so despite getting warmer, we'll need to increase to R-49. It's confusing and unpredictable here.

  8. [email protected] | | #13

    I would like to add continuous insulation above my roof but have some concerns. It is 1911-25 house in Maryland (zone 4). The rafters are true 2 x 6's on 18" centers (!) with board sheathing. Years ago I insulated with fiberglass leaving an airspace and then 1/2"polyiso on the interior face. I have hips and valleys. I am considering adding rigid and a metal roof. Would it be crazy to leave the airspace under the lower sheathing and cap it at the ridge with permeable housewrap so that the assembly has a way to dry?

  9. GBA Editor
    Martin Holladay | | #15

    Q. "Would it be crazy to leave the airspace under the lower sheathing and cap it at the ridge with permeable housewrap so that the assembly has a way to dry?"

    A. No, it would not be crazy. But your decision takes judgement. The thicker the new exterior foam, the safer the assembly will be. During your insulation work, you need to inspect high and low areas of the existing assembly to assess whether there are currently any moisture issues -- dry materials will give you confidence to move forward. Remember, an airtight ceiling is the key to success.

  10. matmarcum | | #18

    Hey, Martin. I'm always a huge fan of any article relating to exterior insulation so, this was a good read. I have a project house (located in 5A) with which I'm open to experiment with copious amounts of exterior insulation - as much as I can or can afford. I've been wanted to do a Larsen truss project so, I'm curious to know; Would it be unwise to run another set of Larsen trusses, or even I-joists, up the roof and fill that cavity with mineral wool to achieve the entirety of my R-value target above the decking? Then, since it appears I'm going to be burning through three deck board levels (1 above rafters, one on top of whatever type of insulation makes the cut, then lastly, another on top of the Chicago'd 2x4s that create the air-vent channels) I won't have any issue making contact with the fastening surface since it'll be the top of the I-joists/furring strips/metal 2xs. Is this too cute by half or a plausible attack plan?

    **Move the ZIP sheathing to the DECKING 1 or DECKING 2 position?

    1. GBA Editor
      Martin Holladay | | #19

      Your proposed roof assembly is expensive and labor-intensive to assemble, but it will provide excellent performance as long as (a) the layer you call "Decking 1" is a robust air barrier, and (b) you install some type of blocking to prevent the I-joists from twisting, and (c) you remember to install roofing underlayment above the layer you call "Decking 3."

      There will be some thermal bridging through the I-joists, but the thermal performance of the roof assembly won't be seriously undermined by this thermal bridging.

      1. matmarcum | | #36

        Hey, Martin - thanks for the feedback. Would it be outrageous to utilize less wood by creating my own trusses with sparse webbing out of cut-in-half 2x4s and 7/16 plywood - like making my own Larsen trusses? If that wouldn't work, what about utilizing metal studs to reduce the cross section of thermal bridging from 7/16 to 1/16? Would that be counterproductive as it would create a cold condensation surface within the assembly? I have some available labor between jobs/cancelled or delayed jobs for which I can make use of salaried guys without additional labor costs, but looking at the price of the rigid foam, I was hoping to make it to R-40 above the rafters and use fluffy stuff in the bays against the underside of the sheathing that rest on the rafters. It appears that there's no free lunch, here. On a spread out ranch home, this'll be somewhere in the ballpark of $17k for the rigid board, alone (625 sheets @ 2"). Ouch.

        1. MartinHolladay | | #37

          Home-made Larsen trusses for use as roof joists may work, but whether or not the design you have in mind will work depends on lots of factors. Before you proceed, talk to an engineer.

  11. williamfinch | | #20

    Martin, thanks for your articles. An element that I'm confused about: In Zone 2 or 3, where reduction of heat gain is the primary objective, if you used two or more layers of above deck isoboard insulation, would you want or need an air barrier on the roof deck? Wouldn't the air barrier be as cost effective and maybe more effective above the isoboard (as in a zip system deck) or integrated into the isoboard itself at any layer (as with taped isoboard seams?). I should add, in the house construction project I'm looking at, the decking above the rafters is attractive 7(!)-quarter oak tongue and groove, so I'm trying to preserve its usefulness and inside appearance.

  12. GBA Editor
    Martin Holladay | | #21

    Climate Zone 3 includes cities like Little Rock, Arkansas and Oklahoma City. It's certainly true that in Zones 2 or 3, there is less chance of damp roof sheathing than in colder climates.

    In my opinion, though, you still want a robust interior air barrier in Zones 2 and 3, because you still want to prevent the migration of humid interior air to the cold roof sheathing on the coldest day of the year. (The coldest historic temperature in Oklahoma City was -17 degrees F, which is quite cold.)

    By "isoboard" insulation, I assume you mean polyisocyanurate rigid foam. If you follow my advice and install a robust air barrier near the interior ceiling, you don't necessarily need any more air barriers elsewhere in the assembly. That said, tape is relatively cheap compared to repairing a damp roof, so I still think it's a good idea to include tape on one of the other layers in your assembly (either rigid foam or the upper layer of OSB or plywood sheathing).

    Leaving the tongue-and-groove ceiling boards exposed is fine, as long as you install a robust air barrier above the boards.

  13. williamfinch | | #22

    Thanks ---I keep looking for an acceptable short way to avoid typing polyisocyanurate rigid foam insulation ... I expect your advice is fail safe. I'll admit, I've kind of ignored concerns about cold temperatures on roof decks in this zone. Will think about it more carefully. With due respect to all the many variables (metal roofs, conditioned and unconditioned spaces, insulation above or below) I suppose the building codes had in mind a temperature (or duration of temperatures) where moisture damage from humidity contact with cold roof becomes more likely? One of the limitations of the zones is that they don't necessarily capture the significant differences in climate within each zone. ln much of the eastern zones 2 and 3, for example, temperatures (barely) below zero are approximately once in a century events. An unusually cold winter day (maybe once or twice a year) in our section of zone 3 would be near 10 degrees, usually with fast recovery to above 32 degrees by mid-morning. But it may be that the effective temperature of concern is closer to 40 degrees?

    1. GBA Editor
      Martin Holladay | | #23

      I get it -- truly cold weather in Climate Zones 2 and 3 is rare. But remember -- once the roof sheathing gets wet, it can't dry easily.

      I tend to be conservative when it comes to designing insulated roof systems. A belt-and-suspenders approach during new construction (or whenever you are insulating the roof) is always cheaper than tearing open a roof in 6 years to make repairs.

      By the way -- the short way to refer to polyisocyanurate is to call it polyiso.

      1. jesjonesy | | #38

        I am looking for a belt and suspenders approach as well in zone 3 California and this updated article has really helped. I still need some air sealing advice however.

        First, I made a huge mistake when I extended my 2x6 t&g exposed roof decking out over the top plates for a roof overhang. I have corrected this hopefully by drilling out and calking the t&g at the outside face of wall, and cutting out and calking the 3/8 in ply sheathing to the 2x6 continious around the exterior wall permiter.

        Assuming this is adequate I am now trying to establish an air barrier above the 3/8" sheathing. I have taped the seams but am concerned about the integrity of the plywood. It rained in California in the summer thanks to my project, and I used 3 ply non exterior rated sheathing which has delaminated in some areas on the first layer, possibly more. The staple holes look gapped a tiny bit as well in some places.

        My question is can I put down an air barrier product like blueskin vp100 or something rated for a very low slope roof while still allowing drying to the inside? My section is then two layers of taped and staggered polyiso, one layer of xps, 1/2" sheathing, peal and stick membrane for low slope roof, and then standing seam metal.

        Thank you!

        1. MartinHolladay | | #39

          Jes Jones,
          You've made two errors. Fixing this will be hard.

          To answer your fist question: No, caulking the seams in cantilevered tongue-and-groove ceiling boards will not make your ceiling airtight. You can expect differential movement due to changes in temperature and humidity that will break the airtightness of the caulk. The best solution would have been to cut off the tongue-and-groove boards at the top plate and to devise another way to create your overhang.

          Second, it sounds as if your 3/8-inch plywood is in terrible shape, with rain-caused delamination. It's not a sound substrate. The best solution is hard to determine without a site visit, but covering the delaminated 3/8-inch plywood with a decent layer of 1/2-inch Zip sheathing would probably be a good idea..

  14. mark_sambol | | #24

    Thank you for the very informative article. I'm in Zone 3 (Georgia) and considering a cathedral ceiling with an insulated assembly very similar to the detail that was included in the article. There's no mention in the article of a vapor diffusion port above the ridge. Is this something that needs to be considered for Zone 3?

    1. GBA Editor
      Martin Holladay | | #25

      The approach described in this article does not require vapor diffusion ports, since using an adequate thickness of continuous exterior rigid foam should prevent the accumulation of moisture in the sheathing. If there is some reason why you're planning to install foam that is thinner than recommended in the guidelines provided in this article, the risk of moisture accumulation increases -- and vapor diffusion ports might be appropriate. (That said, unless the assembly is carefully detailed, such an approach might be risky.)

      For more information on this issue, see "Vapor Diffusion Ports."

  15. tbanwart | | #26

    "Keeping the insulation in contact with the underside of the roof sheathing". I'm planning to insulate our new roof with 2 layers of EPS on top of our decking and then fluffy insulation below. If a person does not want to use Spray foam on the inside, what is best option to keep fluffy stuff against the sheeting? Will blown /glued cellulose work? What about netting? I'm not sure of our rafter sizing yet but I'm afraid the batts will sag and not sure how critical this is. Open cell foam is an option but I'd prefer to avoid and use the fluffy but now I'm concerned about the air gap.

    1. MartinHolladay | | #30

      User 6128827,
      Q. "What is the best option to keep fluffy stuff against the sheathing?"

      A. The answer depends on your framing type and framing thickness. Roof trusses with 2x4 chords will require a different approach from solid sawn rafters.

      If you have roof trusses, I think that open-cell spray foam is probably your best option. If you have solid rafters, you have to do the math. The first step is to calculate the R-value of fluffy insulation (for example, fiberglass batts) that completely fill the rafter bays. If this R-value is too high to meet the ratio rules spelled out in this article, you'll have to increase the thickness of the rigid foam above the sheathing to make the ratio rules work.

      If the R-value of the proposed fluffy insulation is too low, you'll have to scab on extra material to deepen the rafters.

      It's also possible to install ledgers on either side of each rafter to help with the installation of plywood or OSB inserts to keep the fluffy insulation tight to the rafters. But this approach is so labor-intensive and expensive that I don't recommend it.

      1. tbanwart | | #41

        Thanks for your reply. When the installers screw through multiple layers of foam, osb, is there any concern for their holes left behind where the screw misses the main rafter below and they have to remove and try again? I'm guessing this will happen a fair amount when doing a large roof and going through 4" or more of foam. Do these little holes become pathway's for moisture then to come up and condensate? Just curious how big of issue this is.

        1. tbanwart | | #53

          Anybody have some thoughts on this or tips for getting long screws in straight?

          1. Expert Member
            MALCOLM TAYLOR | | #54


            Buy a cheap portable drill press, and attach a larger plywood base to keep it flat to the foam.

  16. peteramerongen | | #27

    Thanks Martin. Great that you are interrupting retirement to pass on your hard earned wisdom.
    We have numerous projects in Climate Zone 7A where the owners have filled their attics ( Typically 4:12 pitch with ~4" heel) with cellulose to about R50 but have done nothing about air tightness. Blower door tests are in the 5 to 8 ACH50 range. There is usually some kind of vapor barrier at the ceiling level.
    Your approach. is very attractive. We could saw off the existing overhang, wrap the existing roof with a vapour open, air tight WRB tied to a new wall WRB, Add 6" of Neopor or equivalent (R30), add new overhang brackets, and then sheath the whole thing with OSB.
    My question is: Is the new R30 over R50 ratio risky? The bottom of the original sheathing would get cold periodically, but with air leakage stopped, the moisture drive would be restricted to diffusion through the gap in the existing leaky ceiling vapour barrier.
    Would diffusion vents at the peak help?
    (Pretty sure we could get structural engineering sign off. Would need to check.)

    1. MartinHolladay | | #29

      According to the ratio rules for roofs, R-30 rigid foam plus R-50 fluffy doesn't work in Zone 7. (You're suggesting that 37% of the total R-value of the roof assembly come from rigid foam, when in fact you need that percentage to be 61% or higher.) But you're breaking another rule, too, since your fluffy insulation isn't in direct contact with the underside of the roof sheathing, as it should be.

      There are lots of potential new issues with your suggested assembly. My gut instinct tells me that everything will probably be fine, but with a leaky ceiling, you might find yourself in trouble. For more information on your suggested approach, see these articles:

      "The History of the Chainsaw Retrofit"

      "A Real Chainsaw Retrofit"

  17. peteramerongen | | #32

    Thanks for this Martin, I'll just up the quantity of rigid to a safe level and forget the diffusion vent. My instinct is that it would work too. We don't have that many heating days when the bottom of the original sheathing is below the dew point. We would get some frost in really cold spells but not so much that the diffusion vent couldn't handle it. Difficult to test.
    Thanks again

  18. tbanwart | | #40

    I had a question and I refer to page:

    1) Instead of rigid foam on exterior, can I do unvented rafter bays by flashing with 2″ or more of closed cell foam, then packed cellulose (or other type), then 2″ of rigid foam under the rafters to stop thermal bridging? This would be variation of assembly #5 in the link above. Is this permissible or would this cause any moisture issues. I’ve looked and saw something like this reference in this article, but wanted to confirm. I'm attaching a quick drawing. Just trying to see if there is a way to use the closed cell foam / batt insulation for air sealing and then rigid foam below to address thermal bridging and give radiant barrier above sheetrock. I understand the rigid foam on top is probably best but in our area, none of the contractors have done this before and they are apprehensive.

    1. tbanwart | | #45

      Anybody have any thoughts on this? Can you do an unvented roof with closed cell foam against sheathing, then fluffy insulation, and then rigid foam over the rafters on inside to hold all the fluffy in and break thermal bridging of rafters? Sort of a foam sandwich with the fluffy in between.

      1. brendanalbano | | #46

        I'd be comfortable with the spray foam, fluffy, rigid insulation sandwich as long as the rigid insulation was vapor permeable. When you are doing your calculations for the proper ratio of spray foam to total insulation, make sure to include the rigid on the "total" side of the ratio.

        In your drawing, is the area above the spray foam going to be vented? It looks like it would need to be.

  19. johnnydc | | #42

    Martin, thanks as always for the clear and helpful write up. This is something I'm pretty determined to do, and your article is incredibly helpful.
    My main concern is: how do you think that this works with the solar arrays? I have solar panels now, and I'm wondering if your experience tells you that there is some helpful way to make sure that when the solar goes back on, the installers don't just punch hundreds of holes into your sheathing and insulation while looking for the rafters. My friend does solar installation for a big company, and his stories do not reassure me about the care and precision with which his colleagues work! Even in the best of circumstances, they miss rafters and have to patch the holes, and if you get somebody who is careless or incompetent, the situation can be worse.
    Your advice would be most appreciated.

  20. bigtoe | | #43

    Hi Martin,
    I'm over in climate zone 6 in the White Mountains of NH (Campton). I'm working on a project with above the code minimums for our climate zone and using the roof assembly you mention in this article. The exterior foam (EPS) vs interior fluffy (Rockwool) percentages are with in your recommended limits. My question is about using the fluffy insulation and avoiding the air gap with the interior sheathing. In the article you mention:

    "An air gap between the sheathing and the fluffy insulation (or the spray foam insulation) is not permitted.

    Complying with this requirement can be tricky: for example, if you are using fiberglass batts, the batts have a tendency to sag away from the roof sheathing, creating an unacceptable gap between the sheathing and the fiberglass. That’s why many builders find it easiest to use open-cell spray foam under the roof sheathing rather than fiberglass, mineral wool, or cellulose."

    Should using the fluffy insulation (fiberglass, mineral wool, or cellulose) in this type of assembly be avoided at all costs? Or given the extra cost of using open cell spray foam, is the fluffy insulation the next best thing but do the best job you can installing it and drying to the inside will just take longer?

    Thanks for all your writings!

    1. MartinHolladay | | #44

      From an environmental perspective, using fluffy insulation like fiberglass, mineral wool, or cellulose is definitely preferable to using open-cell spray foam. You just have to plan ahead to make sure that the insulation is properly encapsulated, so that the fluffy insulation is slightly compressed against the roof sheathing to avoid sagging. That can be tricky, but it's certainly possible.

      1. bigtoe | | #47

        Thanks so much for the reply!

  21. kiwiscott | | #51

    2 Quick Questions.

    1. With the change to R-60 for Zone 5a, and all zones, in 2022 does the 41% change?
    2. Can one build an internal assembly within an existing attic (think a sub roof) that forms this insulation / air barrier.


  22. SSHabal | | #52

    Martin, thanks a lot for your article.

    I'm building a small classic cape. Can you (or other community members) please share your thoughts regarding two questions I have?

    1. “If you plan to install steel roofing, you may be able to skip the upper layer of roof sheathing”. Would 2x4s be needed on top of foam or are 1x4 purlins sufficient? Or both 2x4s to secure foam boards screwed to the rafters and 1x4 purlins parallel to the ridge?

    2. regarding air barrier over sheathing - any feedback regarding comparing ice shield over plywood vs taped zip panels? The cost seems to be about the same.

  23. tbanwart | | #55

    I have a question regarding fastening the 2 layers of foam / 2nd sheathing layer. I'm guessing it is really difficult to screw a 6" screw down perfectly straight and never miss the 2x12 rafter below 3-4" of foam and then the OSB. On the entire roof, any time a screw is missed that will create a hole in the 2 layers of insulation plus the bottom layer of taped OSB / Zip board. Are those holes of "Misses" of concern for moist air coming up into and touching the top layer of OSB or is really a non - factor.

  24. oren_h | | #56

    I am installing a concrete Spanish tile roof and would like to have a layer of rigid foam exterior to my roof sheathing. I’m located in Southern California (climate zone 3B) and only need R-5 on the exterior. My roofing sub told me that polyiso cannot be installed under tile roofs, because it will become damaged during the installation. Instead he recommended a Low-E Thermasheet product which is ¼” thick and has an R-value of 1. But with an airspace which I plan to include, since the tile will be installed over arched battens, the R-value is listed at 6.4. I have read the articles about false claims of foil-faced bubble wrap, which makes me a bit weary. I called the company and they sent me the attached 3rd party testing of the insulation values. They were up front about the need for a reflective air space to achieve the reported R-values. Assuming I include the air space, can this in fact be used as an alternative to rigid foam?

    1. MartinHolladay | | #57

      R&D Services is a reputable lab. Their report is believable.

  25. zrtyler | | #58

    I have been referencing your articles and others quite a bit, they are very helpful. I am in the middle of retrofitting 4.5” polyiso in climate zone 4C above roof deck. Existing roof deck is a patchwork of new 3/4” CDX and old 1x8 “skip lap” boards. I have omitted the air barrier on top of existing sheathing but can have been spray foaming every seam of the polyiso boards. It sounds like this is only the “belt” of “belt and suspenders”, but how concerned would you be with this assembly?

    1. MartinHolladay | | #59

      Z. R.,
      What materials will be installed above the polyiso? If you are installing a layer of sheathing above the polyiso -- that's typical in most situations -- then I advise you to tape the seams of the upper layer of sheathing with a high-quality tape.

      1. zrtyler | | #60

        Sorry I forgot to mention, I’m installing 1/2” CDX then synthetic underlayment/asphalt shingle. I had thought about using sealant on the seams, but I will tape them, thank you!

  26. MikefromtheMountainsofUtah | | #61

    I'm going to follow your lead on this on my build in northern utah. The "stack" for my build is TJI single slope low roof, OSB, Ice watershield per code, Then I'm thinking R20 of polyiso with an inch or so of non-iso above to mitigate the R drop. I was planning on another layer of OSB and then standing seam roofing. Does that sound like a good plan? As I understand it the standing seam should not need furring strips, which would certainly make it safer to walk on without denting the roofing.

    1. Expert Member
      MALCOLM TAYLOR | | #62


      Can you clarify why the code requires the I&WS? It in the right place if used as an air-barrier (although it wouldn't be my first choice), but if it is required as a roof underlay, it needs to be above the top layer of roof sheathing.

      1. MikefromtheMountainsofUtah | | #63

        Good catch. I should have said it was designated by the engineer due to my low slope roof. It makes sense that it would be above the top layer, leading to everything drying inward and not collecting moisture in the polyiso. Thank you

Log in or become a member to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |