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

Insulating a Wood-Framed Floor Assembly

How to insulate a cold floor

This detail shows one way to insulate a cantilevered floor. While the rigid foam at the top of the floor assembly is optional, attention to airtightness is not. The main reason that cantilevered floors feel cold in winter is air leakage. [Image credit: Fine Homebuilding]

Most wood-framed floor assemblies — for example, floor joists over a basement — have conditioned space on all sides, and therefore don’t need insulation. Sometimes, though, a wood-framed floor assembly has outdoor air on the underside — as is the case with a house on piers, for example, or the cantilevered floor of a bump-out, or the floor of a bonus room over a cold garage. In these cases, builders need to do a good job of insulating and air-sealing the floor assembly.

In 2012, I wrote a Fine Homebuilding article on insulating wood-framed floor assemblies. But that article was relatively short, and I haven’t yet written a comparable article for GBA. It’s time to revisit the topic.

Why do we insulate ceilings to a higher R-value than floors?

Building codes require that ceiling assemblies include more insulation than floor assemblies. Many builders explain this requirement by saying, “Heat rises.” In fact, if we’re talking about conduction and radiation — two of the three heat flow mechanisms — heat will flow equally in all directions, including downward, from warm surfaces to cold air. When we consider heat flow associated with air movement (convection), however, the issue gets more complicated. While we can’t accurately say that “heat rises,” it’s certainly true that warm air rises.

In most homes, air leakage and the stack effect ensure that air near the ceiling is warmer than air near the floor. (For more information on these mechanisms, see “Cold Floors and Warm Ceilings.”) That’s one reason that codes require higher R-values for ceilings than floors.

The second reason for the requirement is that it is usually cheaper to install high-R attic insulation than it is to install high-R floor insulation or wall insulation. In other words, the fact that attic insulation is cheap to…

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. nshirai | | #1

    Martin, thanks for this fantastic analysis of these often tricky details. It seems worth mentioning that while the 2012 IECC does forbid putting the insulation at the bottom side of such assemblies, an exception made it into the 2015 IECC that permits it (presumably through Dr. Joe's influence). The same language is carried into the 2018 IECC as well.

    Here's the reference:

    "R402.2.8 Floors. Floor framing-cavity insulation shall be installed to maintain permanent contact with the underside of the subfloor decking.
    EXCEPTION: The floor framing-cavity insulation shall be permitted to be in contact with the topside of sheathing or continuous insulation installed on the bottom side of floor framing where combined with insulation that meets or exceeds the minimum wood frame wall R-value in Table 402.1.2 and that extends from the bottom to the top of all perimeter floor framing members."

    Ultimately what comes out of the inspector's mouth is what governs, but since many jurisdictions are moving to more recent versions of the code this seemed worth mentioning. I'd argue that some additional language emphasizing air sealing when using this approach should have been included, since without it the code relies on builders to apply the general concept of insulation being installed in contact with an air barrier to these assemblies, which seldom happens in real life. I do not bring this up to advocate for one approach or the other, but only to point out that more recent versions of the code permit the alternate method, and hopefully echo your emphasis on air sealing to anyone who happens to read this comment.

    1. GBA Editor
      Martin Holladay | | #2

      I appreciate very much your information on recent versions of the code. Thanks.

  2. Expert Member
    Dana Dorsett | | #3

    >"This detail shows one way to insulate a cantilevered floor. While the rigid foam at the top of the floor assembly is optional, attention to airtightness is not. "

    But the rigid foam at the BOTTOM of the assembly is an actively BAD idea, since it inhibits the ability of the assembly to dry. It's a moisture trap.

    While thermal break the bottom side foam provides on the joists is welcome, that can be achieved with joist-edge strips deepening the cavity and filling the entire cavity with fiber insulation, with a much improved drying capacity toward the exterior/bottom.

    1. GBA Editor
      Martin Holladay | | #4

      I disagree. Moisture accumulation on the exterior side of an insulated floor assembly is extremely unlikely. It's not like moisture accumulation in wall sheathing or roof sheathing -- because the stack effect works in favor of protecting floor assemblies, while also working to raise the risk in roof assemblies.

      I've never heard of a moisture problem associated with too-thin exterior rigid foam used on a floor assembly.

      1. Expert Member
        Dana Dorsett | | #5

        >"Moisture accumulation on the exterior side of an insulated floor assembly is extremely unlikely."

        True, unless the interior air sealing details at the band joist over supporting walls aren't attended to.

        It's still a moisture trap, even if it's a lower-risk moisture trap than it would be in a roof assembly.

      2. MCShaw | | #14

        We recently demolished a 1970's home in Climate Zone 3 that experienced exactly this problem. At some point in its history, the owners covered fiberglass insulated floor joist bays with a continuous layer of rigid foam insulation on the underside of the joists. This caused moisture to accumulate in the bays and completely rot out the joists. The house was not salvageable.

    2. hughw | | #8

      Dana, 5 weeks ago, in another thread on insulating an exposed floor you stated.

      "The vapor permeance of a 3/4" plywood subfloor is pretty low, (and thus protective), but I'd still be reluctant to put thin foil faced goods on the underside of R30 in MV. Unfaced or fiber faced foam works fine, and has at least a comparable vapor permeance to the subfloor, for at least SOME drying capacity toward the exterior."

      You further recommended exploring the use of asphaltic fiberboard insulation...Change of heart? or have I missed something. Currently, I've kind of given up on the fiberboard but was planning to go with 1-1/2" EPS under TJI joists and then cover with 1/2" MDO plywood, all nice and sealed. Bad idea? I know that the TJI's are not too bad regarding thermal bridging, but like the idea of a continuous insulation "blanket" under the exposed floor.

      1. Expert Member
        Dana Dorsett | | #15

        >"You further recommended exploring the use of asphaltic fiberboard insulation...Change of heart? or have I missed something."

        I think maybe you were misunderstanding my objection to the drawing- there is no change of heart. It's exactly the foam board on the bottom, directly above the bottom side exterior sheathing that I'm viewing as a potential problem. Getting rid of it and sheathing with highly permeable

        The foam board that is snugged up to the subfloor in the picture is still OK (but not often necessary).

        >"Currently, I've kind of given up on the fiberboard but was planning to go with 1-1/2" EPS under TJI joists and then cover with 1/2" MDO plywood, all nice and sealed. Bad idea? "

        Yes, it's probably a bad idea, since MDO plywood typically has a vapor permeance of well under a half-perm. CDX plywood or OSB would offer several times the drying capacity of most MDO plywood. This vendor's 3/8" MDO plywood with MDO on one side only tests at 0.3 perms, about 1/10th the vapor permeance of CDX at a moisture content that would support mold.

        Unfaced Type II EPS would have a vapor permeance of 1.0-1.5 perms @ 1.5" which is OK, not great, yet 3-5x the drying capacity of one-side MDO plywood.

    3. brianvarick | | #17

      Would continuous insulation at the bottom work well if there was none against the subfloor?

      1. GBA Editor
        Martin Holladay | | #18

        As you've probably noticed, Dana Dorsett and I don't agree on the dangers of installing a continuous layer of rigid foam on the underside (exterior side) of an exposed floor. While Dana has misgivings (due to his worries about moisture accumulation or condensation), I don't.

        Q. "Would continuous insulation at the bottom work well if there was none against the subfloor?"

        A. I'm not sure what you mean by "work," but in my mind, continuous exterior insulation is desirable -- more desirable than insulation against the subfloor, because the exterior insulation (assuming it is properly installed) actually reduces thermal bridging through the joists, while any insulation against the subfloor does not.

        Dana worries about moisture accumulation against the foam, but there isn't any evidence that that ever happens. For one thing, the OSB or plywood subfloor is a vapor retarder, so there is very little outward vapor diffusion. For another thing, with a floor assembly, there is zero chance that any interior moisture will cause problems by piggyback on exfiltrating air -- a real problem with walls and roofs -- because the stack effect works in your favor with floors. The stack effect is protective for floors but extremely risky for cathedral ceilings.

        One more point: If you install a layer of continuous exterior rigid foam, don't use XPS, since XPS is manufactured with a blowing agent that has a high global warming potential. For more information on this issue, see "Choosing Rigid Foam."

  3. user_8675309 | | #6

    I live in an early 70's split entry in Anchorage, Ak.(zone 7) and have 2 areas which are cantilevered. I few years ago I did a remote wall configuration from the footers to the soffits using 4" of rigid foam(r20) with a layer of either bituthene or 10 mil plastic between the sheathing and the foam. My question is: The Cold Climate Research Facility recommended a 2/3 outside, 1/3 inside insulation. Stuffing 9 1/2" of insulation into the cantilevered bays kind of seems to go against the "rules" . I still have one cantilevered area yet to do. Do a fill it full or do I put r13 in there?

    1. GBA Editor
      Martin Holladay | | #9

      This is a complicated topic, and I don't want to make light of your question. Since you live in Alaska, you need to take the risk of possible moisture accumulation seriously -- more seriously, say, than someone who lives in Chicago.

      That said, Anchorage isn't Fairbanks, and floor assemblies are far less likely to accumulate moisture than wall assemblies or roof assemblies.

      The Cold Climate Housing Research Center is correct -- at least for walls and roofs -- that Alaskans need to be careful when installing exterior rigid foam. Thin foam is definitely more risky and thick foam.

      What would I do in your case? I would probably fill the joist bays with a thick fluffy batt and stop worrying (based on the general principle that moisture accumulation in floor assemblies isn't usually a problem). But I don't have much building experience in Alaska, so I would respect your decision if you decided to be more conservative than me.

      The best (conservative) approach wouldn't be to install an R-13 batt -- it would be to install some closed-cell spray foam against the interior side of your exterior sheathing, followed by a batt that fills the cavity. But R-20 exterior rigid foam plus an R-13 batt would be OK, I suppose -- especially if your air sealing work is impeccable.

  4. Expert Member

    Foam on the underside of exposed floor assemblies is a pretty common approach here. Like Martin, I've never heard of moisture issues. Especially when the floor are close to grade, I'd imagine the moisture drive is primarily from the outside in. I wonder how much the permeability of the floor coverings above affects things?

  5. PAUL KUENN | | #10

    Great article and responses. Just got a second job done this fall with cantilevered floors and bathrooms above. One tiled floor and the other vinyl. One had copper split in the weeks of sub zero and they fixed it from the inside temporarily. Of course, they called me in to ask how it happened. When I opened up the exterior I showed them all the long term moisture damage and rot from air infiltration leading to ice formations under their cold floor. After the ants had fallen on me, I decided to go with the top drawing idea (I must have seen it awhile ago on GBA). EPS under the sub flooring (recently replaced), vertical EPS against the inside of the rim joist, then fiberglass batt, then EPS again at bottom (between joists) and all sealed along the edges with foam. I topped it off (or should I say I bottomed it off?) with leftover Mento Plus from 475. Not quite done. The next week I asked the owners if they didn't mind a little different look. I then covered the complete bottom with a large sheet of 1.5" EPS (so I could easily use 2x2"s to frame out the new exterior) to cover joists. Afterwards I added new fascia and trim to cover the edges. They were fine with it. Last week they called during the freak snow storm to say how wonderful it was to step out of a shower onto a warm floor. Added warmth to my heart to hear that.

    1. GBA Editor
      Martin Holladay | | #11

      Thanks for sharing your story. The homeowners were lucky (or smart) to hire you.

  6. PAUL KUENN | | #12

    I forgot I had shot a pic of the "almost done" before I decided to add a lower sheet of EPS. Certainly looked better.

  7. synergytodd | | #13

    It is my belief that insulation should be in complete and constant contact with the air barrier. In Alabama, most insulators install an R19 batt in bonus room floors or cantilever by resting the batt on the lip of the I-joist that is about 16" deep. There is no air sealing done at all. The only way we insulate bonus room floors/cantilevers is with a minimum of 6" of open cell foam.. We find that with I-joist, there is little thermal bridging when open cell foam is used.

    1. Expert Member
      Dana Dorsett | | #16

      >"The only way we insulate bonus room floors/cantilevers is with a minimum of 6" of open cell foam.. We find that with I-joist, there is little thermal bridging when open cell foam is used."

      Thermal bridging issues aside, 6" of open cell polyurethane is an embodied carbon disaster compared to 6" of damp-sprayed cellulose of equal performance (and a negative carbon footprint.)

      A full depth installation of dry blown cellulose (which would be better still) can sometimes come in cheaper than 6" damp sprayed with lesser I-joist depths, but probably not with 16" I-joists.

      The batt insulated open-web truss middle picture looks like R50 or more, but with the voids & gaps the imperfect fit it's unlikely to be doing nearly that well. With a blown fiber insulation approach the fit would be as close to perfect as it gets in practical construction.

      To be sure, in Alabama there aren't any floor temperature striping issues to speak of compared to locations that go for days without breaking 0F during cold snaps even with milled lumber joists, and installing foam board under cantilevered joists would buy very little in comfort or efficiency at R19 or higher in the cavity.

      1. synergytodd | | #28

        I have been in the cellulose manufacturing and installation business most of my life and there is not a good way to install 6" of cellulose under a bonus room floor on 16" joists without it sagging. I would like to see an example of how this is done?
        I pumped my 16'" I-joist full of cellulose insulation after the sheetrock was installed on my last house. There is almost know way to prevent the material form settling and the cost of material and the amount of labor was prohibitive.

        1. jonscholl | | #37

          Todd I will be building a sunroom addition on helical piles (piers) and plan to install the 2x12 floor joists with dense pack cellulose. Are you saying that this will settle over time and not stay pack in against the underside of the sub-flooring?

          1. Expert Member
            MALCOLM TAYLOR | | #38


            "pump(ing) my 16" I joists full of cellulose" is a pretty vague way of describing what work was done. Properly dense-packed, the cellulose should not settle - and if it does, leaving a small void against the underside of the sub-floor isn't a problem. Joe Lstiburek even recommended it as way to slightly warm the floor.

            The only area where that possibility would cause a problem is at the rim-joists, where perhaps it would be a good idea to use foam board.

  8. Dennis_the_Menace | | #19

    I think I have a scenario somewhat similar to what this article addresses but different, and I'm not sure what to do for insulation. You can see from the image that there is a garage slab against part of the rim joist (with Ice and Rain membrane in-between). In the winter the slab will be quite cold (Climate zone 5) so that the rim joist will be cold as well and may have condensation. I'm thinking that not much moisture will come through the Ice and Rain membrane but that sources of moisture would be (1) up through the foundation concrete and mudsill and also (2) vapor from the basement air. If this is true, then if I insulate the rim joist area and seal it to keep out basement air, then I am trapping moisture from the foundation. Or maybe because there is so much gravel backfill, the foundation concrete will not be a moisture source. If I don't seal it the vapor from the basement air gets in. It seems to me like there is no way to prevent moisture. So it also seems that whatever insulation method I use needs to be vapor open so that whatever moisture does get in there has a chance to eventually dry out. Any suggestions on how to handle this insulation issue?

    1. GBA Editor
      Martin Holladay | | #20

      Your question has nothing to do with the topic of this article.

      That said, the usual advice for insulating rim joists can be found here: "Insulating Rim Joists."

      Moisture accumulation in the rim joist is usually not a problem, as long as (a) the rim joist is dry on the day it is insulated, and (b) your house has a capillary break (for example, foam sill-seal) between the top of the foundation wall and the mudsill. (Note that this capillary break is a code requirement.)

      Older homes may lack a capillary break between the foundation wall and the mudsill. If you have that kind of home, the risk of moisture accumulation in your rim joist is increased. You might consider using EPS instead of polyiso to insulate your rim joist in that case -- because EPS allows a little bit of inward drying. One factor in your favor: your rim joist can dry outward, to a limited extent, because the upper half of the rim joist has no Ice & Water Shield.

      One final point: Installing a basement slab against a wooden rim joist was a mistake. It's a hard mistake to correct now that it's already happened.

  9. user-6955756 | | #21

    Why not just use rock wool and avoid all the concerns about trapped moisture etc?

    1. GBA Editor
      Martin Holladay | | #22

      Mineral wool batts can certainly be used to fill the joist bays. If you do that, however, you haven't addressed thermal bridging through the joists. A continuous layer of rigid foam helps with the thermal bridging problem.

      Perhaps you are thinking of using some of the denser panels of semi-rigid mineral wool in a continuous layer on the exterior side of the joists. That's possible, but a little more difficult to install than rigid foam. With semi-rigid mineral wool, you don't get an air barrier and you can't tape the seams. Moreover, the mineral wool is squishy, which may complicate the installation of the OSB or soffit material used to protect the insulation from critters.

      You are correct that mineral wool is vapor-permeable, so it allows moisture to diffuse through it. That's often an advantage.

      1. user-6955756 | | #23

        Ah, I see that makes sense. I've got a similar detail that I'll be working through in a year or two on the underside of the cabin I'm building. It's up on stilts so the entire floor is cantilevered. My schedule didn't allow me to insulate while I had easy access to the floor so I'll be coming back to it and working up from below unfortunately.

  10. Deleted | | #24


  11. [email protected] | | #25

    Referring to the illustration at the head of the article: why not install foam board on the vertical rim joist as Paul Kuenn did? Is the rim joist of a basement (foam board is recommended from different from a rim joist of a bonus room like a space above a garage? I assume it is the proximity to the ground which bring moisture potential, but a space above a garage is 10 feet from the ground.

  12. GBA Editor
    Martin Holladay | | #26

    The rim joist in the illustration is well insulated on the interior by what looks like 12 or 14 inches of fiberglass -- call it R-42. Moreover, the fiberglass is not in contact with humid indoor air; it's separated from indoor air by at least two air barriers: the 2x10 blocking, which is caulked, and the drywall on the underside of the joists (not shown in the illustration). So everything is good.

    If you prefer rigid foam to fiberglass, rigid foam will work, too.

    1. [email protected] | | #27

      I see now - it's about the total r value of that cavity in a horizontal perspective. I will apply the same principle for the 25' length of wall where I will caulk all seams of the rim joist and measure the space to apply adequate insulation.

      Thank you for the clarification.

  13. jakedouglas | | #29


    I've seen in this article/discussion and others your emphasis on having fluffy insulation between floor joists be airtight on all six sides. This makes sense for best performance, but I don't know how to square it with what seems to be considered fine practice of installing open fluffy insulation onto the floor of a vented attic.

    Assuming an airtight attic floor, won't even blown insulation installed against baffles still experience convective losses starting at the attic floor and escaping upward through the material? Is there a reason we don't, for instance, lay a cheap air barrier such as Tyvek over the top of our fluffy attic insulation to partially mitigate these losses (taking care not to compress it, of course)?

    Do physics or observational data tell us much about the relative importance of the airtightness of insulation in unconditioned attic and floor assemblies, aside from the obviously critical plane facing the conditioned space?


    1. Expert Member
      MALCOLM TAYLOR | | #30


      It's something I'm uncertain about too. I'd find a blog discussing the relative effects of both wind-washing and convective loss on batt insulation really useful.

      1. jlipkowitz | | #39

        Hi Malcolm,

        I've very much appreciated your insightful answers to my questions on the forum. I'm on to insulting the floor of a 490 sq ft round cabin on piers in Maine. My floor joists are 2x6's and the joist bays are wedge shaped, emanating out from a central ring towards to outside wall panels. Essentially the bays are a few inches wide at the center of the structure and 26 inches wide at the exterior, forming pie shape wedges. I will be wrapping the underside of the joists with a inch of EPS, sheathing with plywood and taping seams. My question is what to fill the bays with. I could do one of the following,
        1. punch holes, from the bottom, blow in cellulose, and then plug holes
        2. Apply netting to the tops of the floor joists and blow in from the top
        3. Cut mineral wool bats and install them from the top.

        Concerns on number 1 are not being able to verify the coverage. Concerns on the mineral wool is cutting those wedge shapes accurately and not winding up with uninsulated areas or bunched up material.

        All things equal the mineral wool seems like it could be easier for the home installer, but perhaps the performance would be lower?

        I'd greatly appreciate you or anyone else's opinion between these two choices for insulating between the bays. Thanks so much for your time.

        1. Expert Member
          MALCOLM TAYLOR | | #40


          This is something I'm thinking about right now too. We are just getting the piers poured for a small recording studio I'm building this summer.

          I'd worry about voids with blown in insulation from the bottom - especially with joist bays that get very narrow.

          The problem with working from the top - either blowing cellulose, or using batts - is sequencing. From that point on you have to keep the subfloor dry or the installati0n will be ruined.

          I've gone with slightly thicker foam (1 1/2"), so the cavity insulation isn't as important. We will probably use high density fiberglass batts and try to keep things dry. They are fairly easy to compress, so you can cut them a bit large, and they will fill irregular shapes.

          1. jlipkowitz | | #41

            Hi Malcolm,
            Thanks for your thoughts. I should have specified that the roof is and walls are already on so the floor getting wet isn't a problem. The way the cabin was designed, the walls don't land on a fixed part of the subfloor meaning I can still access the top side of the joist bays and then put the subfloor down.

            Given that, would you be more favorable towards blow in or batts?

            If using the batts, would the high density fiber glass be better in this type of bay with the irregular shape due to the compressabilty than mineral wool? Thanks again,


          2. Expert Member
            MALCOLM TAYLOR | | #42


            I don't think the benefits of blown-cellulose are worth the complexity of installing it on such a small floor - especially into irregular joist bays. HD fibreglass batts would be my pick.

    2. GBA Editor
      Martin Holladay | | #31

      Jack and Malcolm,
      There is definitely some undesirable heat loss (due to convection) from attic insulation that lacks a top-side air barrier. A top-side air barrier would improve attic insulation performance. Experience shows, however, that installing Tyvek in an airtight manner (with taped seams) on top of blown-in attic insulation is (a) awkward bordering on impossible, depending on attic access and head room, and (b) more expensive than the widely accepted alternative of just blowing a few more inches of insulation on the attic floor. Investing $100 in extra insulation gets you more bang for your buck (better performance) than investing $200 in Tyvek and labor.

    3. charlie_sullivan | | #32

      Here's a blog on Energy Vanguard about that issue, which used to be serious problem with fiberglass, but is supposedly solved by better preparation of fiberglass for blown-in application.

      Cellulose resists convection more than fiberglass, so once strategy, particularly for old fiberglass that might be before the improvements Allison discusses, is to blow a top layer of cellulose over the fiberglass.

      1. jakedouglas | | #33

        My interpretation of the article is that those insulation types at least have convection, if any, baked into the R-value rating of a given thickness. It remains unclear to me though to what degree it occurs, and how much would be gained by applying an air barrier to the bottom of fluffy insulation installed against an airtight subfloor, or fluffy insulation installed against an airtight attic floor (however impractical).

        If a fiberglass batt performs to its rated R-value *without* being fully airtight, as installed in an attic, does that mean that a fiberglass batt performs *above* its rated R-value in an airtight wall cavity?

        "The further experimentation they did was to put a covering layer over the top of the loose-fill fiberglass. The two they tried were (1) a polyethylene film and fiberglass blanket combination and (2) R-19 fiberglass batts. Both eliminated the convection and the reduction in R-value."

        It sounds like both coverings included additional insulation rather than just an air barrier, which is puzzling.

        1. Expert Member
          MALCOLM TAYLOR | | #34


          There is some data showing the diminution of R values for various insulation types in this study (tables 8 and 9). My take is that if the losses for convection are similar to those caused by wind-washing, it isn't worth spending much time on.

        2. charlie_sullivan | | #35

          I believe that the effect is a lot more minor for a horizontal attic floor than for a vertical wall, where the height of a convective path through the insulation is much greater. That explains some of what it's recommended in walls more vigorously than for attic floors. That, and the fact that in an attic, you can simply add more insulation and there little incentive to get the best insulation value out of a limited thickness.

          1. Expert Member
            MALCOLM TAYLOR | | #36


            That's what interests me. What is "minor"? Is it in fact worth worrying about in either case for an average builder, or is this something that only needs to come into play in the ratified world of very high performance Passive House type projects. Maybe this is just another one of the things (like framing ) we should expect to de-rate a wall or roof assembly for and move on?

Log in or become a member to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |