Conversion calculator

R-Value Converter

Imperial R and metric RSI are the same quantity wearing the same letter, and they differ by 5.678. Two more habits — adding U-values, and trusting the number on the pack — cost more than that.

Calculator

Try:

21 imperial R

R-21.0 · RSI 3.70

U-value 0.270 W/m²·K, or 0.0476 BTU/h·ft²·°F. One RSI is 5.678 imperial R — the same quantity, not the same number.

Why “R-5” is ambiguous

The same numeral read as each unit
If “21” meansImperial RMetric RSIU (W/m²·K)
Imperial R21.03.700.270
Metric RSI119.221.000.048

A factor of 5.678 between two readings of the same figure. A European board sold as “R 5” is imperial R-28; a US batt sold as “R-19” is RSI 3.35. The factor is exact rather than measured — 3600 seconds, an exactly-defined square foot, five ninths of a kelvin and the International Table BTU.

A resistance is a thickness of a material

150 mm of Mineral wool battRSI 4.29R-24.3
Thickness needed to reach RSI 3.70 (R-21.0)
Materialλ (W/m·K)Thickness
PIR / polyisocyanurate board0.02281 mm
Extruded polystyrene (XPS)0.033122 mm
Mineral wool batt0.035129 mm
Glass wool batt0.04148 mm
Expanded polystyrene (EPS)0.038141 mm
Blown cellulose0.04148 mm

Conductivity is the material property; resistance is what a thickness of it achieves. So “the R-value of mineral wool” is not a number — the question needs a thickness, and the same target takes 81 mm of PIR or 148 mm of glass wool.

Layers add as resistances, never as U-values

A timber-frame wall, layer by layer
LayerRSIIts own U
Plasterboard, 12.5 mm0.0812.50
Mineral wool between studs3.700.27
OSB sheathing, 11 mm0.119.09
Cladding, cavity and surface films0.303.33
Total4.190.239

Add the resistances and invert once: U = 0.239. Add the U-values in the third column instead and the answer is 25.19 106 times too large, and it looks like arithmetic the whole way. Resistances in series add; conductances do not.

The pack number is not the wall number

Cavity insulation aloneRSI 3.70R-21.0 — the number on the pack
With 23% timber through itRSI 2.51R-14.332% of the insulation lost
Heat loss through 30 m² at 20 K162 W239 W once the framing is counted

Timber conducts about four times as readily as mineral wool, so studs, plates and headers form a path around the insulation. A real 2×6 wall at 16-inch centres is roughly 23% timber once corners and openings are counted — well above the 15% a naive stud count suggests — and the assembly performs about a third below the number on the pack. Continuous external insulation exists precisely to break that path.

Each extra R buys less than the last

Attic levels, and what each step actually removes
LevelRSICuts remaining loss byLoss vs R-13
R-132.29100%
R-193.3532%68%
R-305.2837%43%
R-386.6921%34%
R-498.6322%27%
R-6010.5718%22%

Heat loss goes as 1/R, so the saving from a step is 1 − R₁/R₂. Going R-10 to R-20 removes half the loss; R-30 to R-40 removes a quarter for the same additional material. That is the whole reason building codes stop where they do rather than climbing indefinitely — and the reason the first hundred millimetres of loft insulation is worth far more than the second.

What this converter covers

R, RSI and U in both systems, the thickness each material needs to reach a target, and what timber framing gives back.

  • Imperial R-value to metric RSI, and U-value in both systems
  • Thickness of PIR, XPS, mineral wool, glass wool, EPS or cellulose for a target
  • Layers summed correctly, with the U-adding error shown beside it
  • Effective assembly resistance once framing is counted
  • Heat loss in watts, and what each step up in R actually removes
R and RSI U does not add Thermal bridging Real heat loss

Free, no signup — exact by definition, not an estimate.

Updated 8 September 2026

At a glance

Formula shown
RSI = R × 0.176110 · U = 1 ÷ R · RSI = thickness ÷ λ · U_eff = f/R_stud + (1−f)/R_cavity
Scenario support
A metric R 5 board is imperial R-28 · R-21 batts in a 2×6 wall perform at R-14
Educational estimate
Planning support from the values you enter — not professional advice.

One letter, two units, a factor of 5.678

Thermal resistance has two units in common use, and both get written with the letter R. The imperial R-value, in h·ft²·°F/BTU, is what North American insulation is labelled with. RSI, in m²·K/W, is what the rest of the world uses — and metric products are frequently sold as “R 5” or “R 2.7” with no unit stated at all.

Common insulation levels in both units
Imperial RRSIU (W/m²·K)Typically
R-132.290.437A 2×4 cavity
R-193.350.299A 2×6 cavity
R-28.45.000.200A metric “R 5” board
R-305.280.189Warm-climate attic
R-498.630.116Cold-climate attic
R-6010.570.095The practical ceiling

Row three is the trap. A European board marked “R 5” is imperial R-28 — better than almost any North American wall cavity — while a US batt marked “R-19” is RSI 3.35, which a European specifier reading it as RSI would take for five and a half times its real performance. The numbers are not comparable and the letter does not distinguish them.

The factor is exact rather than measured, which is worth knowing because it means there is no tolerance to argue about. An hour is 3600 seconds, a foot is exactly 0.3048 m, a Fahrenheit degree is exactly five ninths of a kelvin and the International Table BTU is exactly 1055.05585262 J. Multiply those out and one imperial R is 0.176110184 RSI.

A quick sanity check worth carrying: RSI ≈ R ÷ 5.7, and R ≈ RSI × 5.7. If a figure looks plausible under both readings, it needs its unit confirmed before anything is ordered.

U-values do not add

A wall is a stack of layers, and the whole is worked out by adding their resistances and inverting once at the end. It is easy to state and easy to get backwards, and getting it backwards produces a number that looks like arithmetic.

A timber-frame wall, done both ways
LayerRSIIts own U
Plasterboard, 12.5 mm0.0812.50
Mineral wool between studs3.700.27
OSB sheathing, 11 mm0.119.09
Cladding, cavity and surface films0.303.33
Correct: sum RSI, then invert4.190.239
Wrong: sum the U-values25.19

The wrong answer is 105 times the right one, and nothing about it looks obviously absurd to someone who has not seen the correct figure. Both methods produce a small number from a column of numbers.

The reason is the same as in electricity, where resistances in series add and conductances do not. Heat crossing a wall passes through every layer in turn, so each layer’s resistance is felt in full. Conductance is the reciprocal of that, and reciprocals do not sum.

Two practical consequences. Thin layers barely matter: plasterboard and OSB together contribute 0.19 out of 4.19, about 4.5% of the wall. And surface resistances — the thin films of still air on each face — are worth about 0.17 RSI on a wall and are not optional in a compliance calculation, even though they look like a rounding error.

The pack number is not the wall number

Insulation is sold with a resistance printed on it, and a wall built with it does not achieve that resistance. Timber runs through the insulation from inside to outside, and timber conducts about four times as readily as mineral wool.

R-21 batts in a 2×6 wall, at different framing fractions
FramingEffective RInsulation lost
0%R-21.0
10%R-17.417%
15%R-16.124%
23% (realistic)R-14.332%
30%R-13.038%

A realistic 2×6 wall at 16-inch centres is about 23% timber once top and bottom plates, headers over openings, corners and partition intersections are counted. A naive stud count suggests 15%, which is one of the reasons predicted performance and measured performance so often disagree.

The arithmetic matters as much as the number. Effective resistance is not the area-weighted average of the two resistances — it is the reciprocal of the area-weighted average of the two conductances. The naive average gives R-17.8 where the correct method gives R-14.3, so the wrong method is optimistic by a further 24%. Heat takes the easy path, and averaging resistances quietly assumes it does not.

This is the entire argument for continuous external insulation. A layer of board outside the framing is not bypassed by anything, so its resistance is added in full to the whole assembly. Fifty millimetres of PIR outside a 2×6 wall does more than thickening the cavity does, because the cavity insulation is already competing with the studs and the board is not.

R belongs to a thickness, not a material

“What is the R-value of mineral wool?” has no answer. The property of the material is its thermal conductivity, λ, in watts per metre-kelvin. Resistance is thickness divided by λ, so it is a property of the piece rather than of the substance.

Thickness needed for RSI 5 (imperial R-28)
Materialλ (W/m·K)Thickness
PIR board0.022110 mm
XPS0.033165 mm
Mineral wool0.035175 mm
EPS0.038190 mm
Glass wool0.040200 mm
Softwood0.130650 mm
Dense concrete1.4007 m

The last two rows are the useful ones. Timber is a poor insulator by the standards of insulation and a good one by the standards of masonry — which is why a stud is a thermal bridge through a batt and a timber frame is nonetheless warmer than a concrete one. And no thickness of dense concrete anyone would build is an insulator at all; seven metres of it matches 110 mm of PIR.

This also explains why a product comparison must be at equal resistance rather than equal thickness. A 100 mm PIR board and a 100 mm mineral wool batt are not alternatives — the board is RSI 4.55 and the batt RSI 2.86. Comparing them by thickness or by price per square metre compares different things.

The λ figures above are typical for the material class. A real calculation uses the manufacturer’s declared value for the specific product, which varies with density and, for foil-faced foams, is declared as an aged figure because the blowing agent diffuses out over years.

Where the extra insulation stops paying

Heat loss goes as 1/R, so each additional unit of resistance removes a smaller share of what is left. Equal steps in R are not equal steps in saving, and that is not a subtle effect.

What each equal step actually removes
StepCuts remaining loss byLoss against R-10
R-10 → R-2050%50%
R-20 → R-3033%33%
R-30 → R-4025%25%
R-40 → R-5020%20%
R-50 → R-6017%17%

The first ten points of R remove half the heat loss. The sixth ten remove a sixth of what is left, which is about 3% of the original. That is why building codes settle at R-49 or R-60 for attics rather than climbing indefinitely, and why the first hundred millimetres of loft insulation is worth several times the second.

Two caveats keep this from being the whole story. Adding insulation to the easiest element first is usually wrong — a house with an R-49 attic and single glazing loses far more through the windows, and the arithmetic above applies to each element separately. And loss through an element is only part of a heat balance: air leakage often exceeds conducted loss in an older building, and no amount of insulation addresses it.

What the table does support is a sense of proportion. Going from nothing to something is transformative; going from good to slightly better is not, and the money is usually better spent on whichever element is currently worst.

Related calculators

Other building and thermal tools:

TemperatureCelsius, Fahrenheit, kelvin and Rankine — and temperature CHANGES, which convert differently from readings.
EnergyJoules, kilojoules, calories, food Calories, kWh, BTU and therms — with the two calories listed apart, since one is a thousand of the other.
PowerWatts, kilowatts, horsepower and BTU per hour — with mechanical and metric horsepower listed apart, since they differ by 1.4% under one word.
AreaSquare feet, square metres, acres and hectares, with the factors squared for you — a square metre is 10.76 sq ft, not 3.28.
LengthMillimetres to miles on the exact 1959 factors, with the mil kept clearly apart from the millimetre — they differ 25-fold.
Densitykg/m³, lb/ft³, lb/in³ and lb/gal, with specific gravity against any water reference and the API scale that runs backwards.

More in Conversion, or browse all calculators.

Sources and methodology

The series and parallel-path methods here are the ones in ISO 6946 — resistances summed, transmittance taken as the reciprocal of the total, and repeating thermal bridges handled by combining conductances over area rather than averaging resistances. The distinction between conductivity and resistance follows ASTM C168. The conversion factor is exact because everything in it is defined: the foot, the Fahrenheit degree, the hour and the International Table BTU.

Conversion note

This converts units and illustrates the arithmetic; it is not a building thermal calculation. A compliance U-value uses the manufacturer's declared lambda for the actual product, the correct surface resistances for the element and its orientation, air layers treated to the standard, and a proper treatment of every repeating and non-repeating thermal bridge — lintels, jambs, sills, junctions and fixings — which this does not model. The conductivity figures here are representative of a material class rather than measurements of any product, and real lambda varies with density, moisture, temperature and age; foil-faced boards in particular are declared with an aged value. The framing fraction is an assumption about your construction, not a measurement of it. Insulation levels quoted are illustrative rather than the requirement anywhere: codes differ by country and climate zone and are revised regularly. Condensation risk is a separate calculation that adding insulation can change, sometimes for the worse. Where the answer has to satisfy a building code, use the software and the declared product data your jurisdiction accepts.

How we calculate · Found an error? email us

Authorship & verification

Written and maintained by , a business operator who builds spreadsheet-based calculators.

What's changed (12 updates)

Published 8 September 2026

  1. Published the R-Value Converter: imperial R-value, metric RSI and U-value in both systems, with the thickness each insulation material needs to reach a target.
  2. Establishes that imperial R and metric RSI are the same quantity written with the same letter and differing by 5.678, so a European board sold as R 5 is imperial R-28 while a US batt marked R-19 is RSI 3.35.
  3. States the conversion factor as exact rather than measured, deriving it from four definitions -- 3600 seconds, an exactly defined square foot, five ninths of a kelvin, and the International Table BTU.
  4. Shows that U-values do not add: summing them for a real timber-frame wall gives 25.19 where the correct method gives 0.239, a number 105 times too large that looks like arithmetic the whole way.
  5. Explains why -- heat crosses every layer in turn, so resistances in series add and their reciprocals do not -- and notes the consequences, that thin layers contribute about 4.5 percent of a wall while the surface air films are not optional.
  6. Separates the number on an insulation pack from the number the assembly achieves: timber conducts about four times as readily as mineral wool, so R-21 batts in a realistic 2x6 wall at 23 percent framing perform at about R-14.
  7. Uses the parallel-path method rather than an area-weighted average of resistances, and reports the difference: averaging resistances gives R-17.8 where the correct method gives R-14.3, an optimism of a further 24 percent.
  8. Makes the case for continuous external insulation from that arithmetic -- a board outside the framing is not bypassed, so its resistance is added in full.
  9. Establishes that R is not a material property at all: conductivity is, and resistance is a thickness divided by it, so RSI 5 needs 110 mm of PIR or 200 mm of glass wool and a product comparison must be at equal resistance rather than equal thickness.
  10. Shows diminishing returns concretely -- R-10 to R-20 removes half the heat loss, R-30 to R-40 a quarter, R-50 to R-60 a sixth -- which is why codes settle rather than climbing, while cautioning that the arithmetic applies per element and that air leakage often exceeds conducted loss.
  11. Names what the page is not: a compliance calculation needs declared product lambda, correct surface resistances, every non-repeating thermal bridge, and a separate condensation risk assessment that adding insulation can change.
  12. Verified by 60 automated cases, asserting the conversion factor against its four-definition derivation, that the parallel path always lies strictly between the framing and cavity values across a hundred framing fractions, and that equal steps in R buy strictly less each time.

Add this calculator to your site

Responsive embed — and private: nothing your visitors type leaves their browser.