What psi should I set my keg to?
For the number most people are after — a typical ale at 2.5 volumes of CO₂, beer at 38 °F in a kegerator at sea level — it's 11.2 psi, 0.78 bar on a metric regulator. That is the equilibrium: the pressure at which CO₂ dissolving into the beer and CO₂ leaving it are equal, with 2.5 liters of gas per liter of beer in solution. Henry's law says that equilibrium depends on two things, the temperature and the pressure, and every carbonation chart ever printed is that one relation tabulated. This page uses the standard regression of the ASBC solubility data:
psi = −16.70 − 0.0101·T + 0.001165·T² + 0.1734·T·V + 4.243·V − 0.0684·V² · T in °F, V in volumes
Against the equilibrium table the Brewers Association prints in its draught manual — ASBC data, 33 to 42 °F, 2.1 to 3.1 volumes — the fit lands within 0.4 psi of every cell and within 0.2 psi of 95 of the 110, and their headline example, 11.3 psig for 2.5 volumes at 38 °F, comes out at 11.2. A regulator's gauge is the coarser instrument by a wide margin. What the formula does not contain is a clock, and the second half of this page is about that.
Which temperature — the fridge, the room, or the beer?
The beer's. The chart is a property of the liquid, and the liquid lags the air: a 5-gallon keg put into a 38 °F fridge at cellar temperature is still in the 50s the next morning. Set the chart's 11.2 psi on it that evening and this is what the gas is actually doing, at the same regulator setting, as the beer cools:
| Beer actually at | 12 psi settles at | 2.5 volumes would need |
|---|---|---|
| 34 °F · 1 °C | 2.78 vol | 9.2 psi |
| 38 °F · 3 °C | 2.57 vol | 11.2 psi |
| 42 °F · 6 °C | 2.38 vol | 13.3 psi |
| 50 °F · 10 °C | 2.06 vol | 17.6 psi |
| 60 °F · 16 °C | 1.73 vol — flat | 23.1 psi |
| 70 °F · 21 °C | 1.46 vol — flat | 28.8 psi |
Sea level. Same regulator, different beer temperature. Computed by this page's engine.
A 60 °F keg on 12 psi is heading for 1.7 volumes, not 2.5 — a flat beer that the gauge insists is fine. It isn't a disaster, because the beer keeps cooling and the equilibrium moves with it; it just means the "two weeks" everyone quotes starts when the beer is cold, not when it goes in. The reverse mistake matters more: carbonating warm on purpose, at the 23 psi that holds 2.5 volumes in 60 °F beer, and then chilling the keg without turning the regulator down. Cold beer at 23 psi keeps taking on gas until it's a foam cannon. The temperature swing inside an ordinary fridge is worth knowing too: each degree Fahrenheit moves the chart by about 0.5 psi, so a fridge that cycles between 38 and 42 °F is really asking for two different pressures 2.1 psi apart. Set for the warmer end and you'll never over-carbonate.
One "volume" is a liter of CO₂ (at 0 °C and 1 atm) dissolved in a liter of beer — about 1.96 g of gas per liter, or 2 g/L if the can says so. It's a ratio, so it doesn't change between gallons and liters, which is why a style guide written in Munich and one written in Denver agree on 2.5. The pressure that holds it does change with altitude, and that's the next section.
Force carbonation chart (printable)
Regulator psi for beer temperature down the side and target volumes across, at sea level. The strip underneath is what to add at altitude. Metric regulators: 1 bar is 14.5 psi, so divide by 14.5 — or use the calculator, which prints both.
| Beer temp | 1.5 | 1.8 | 2.0 | 2.2 | 2.4 | 2.5 | 2.6 | 2.8 | 3.0 | 3.2 | 3.5 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 32 °F 0 °C | — | 1.6 | 3.5 | 5.4 | 7.3 | 8.2 | 9.2 | 11 | 12.9 | 14.8 | 17.6 |
| 34 °F 1.1 °C | — | 2.3 | 4.3 | 6.3 | 8.2 | 9.2 | 10.2 | 12.1 | 14.1 | 16 | 18.9 |
| 36 °F 2.2 °C | 0 | 3.1 | 5.1 | 7.2 | 9.2 | 10.2 | 11.2 | 13.3 | 15.3 | 17.3 | 20.3 |
| 38 °F 3.3 °C | 0.7 | 3.9 | 6 | 8.1 | 10.2 | 11.2 | 12.3 | 14.4 | 16.5 | 18.6 | 21.7 |
| 40 °F 4.4 °C | 1.4 | 4.7 | 6.8 | 9 | 11.2 | 12.3 | 13.4 | 15.5 | 17.7 | 19.8 | 23 |
| 42 °F 5.6 °C | 2.1 | 5.5 | 7.7 | 10 | 12.2 | 13.3 | 14.4 | 16.7 | 18.9 | 21.1 | 24.4 |
| 44 °F 6.7 °C | 2.8 | 6.3 | 8.6 | 10.9 | 13.2 | 14.4 | 15.5 | 17.8 | 20.1 | 22.4 | 25.8 |
| 46 °F 7.8 °C | 3.5 | 7.1 | 9.5 | 11.8 | 14.2 | 15.4 | 16.6 | 19 | 21.3 | 23.7 | 27.2 |
| 48 °F 8.9 °C | 4.2 | 7.9 | 10.4 | 12.8 | 15.3 | 16.5 | 17.7 | 20.1 | 22.6 | 25 | 28.6 |
| 50 °F 10 °C | 4.9 | 8.7 | 11.3 | 13.8 | 16.3 | 17.6 | 18.8 | 21.3 | 23.8 | 26.3 | 30.1 |
| 55 °F 12.8 °C | 6.8 | 10.8 | 13.5 | 16.2 | 18.9 | 20.3 | 21.6 | 24.3 | 27 | 29.7 | 33.7 |
| 60 °F 15.6 °C | 8.7 | 13 | 15.9 | 18.8 | 21.6 | 23.1 | 24.5 | 27.4 | 30.2 | 33 | 37.3 |
| Elevation | 1,000 ft 305 m | 2,000 ft 610 m | 3,000 ft 914 m | 4,000 ft 1,219 m | 5,000 ft 1,524 m | 6,000 ft 1,829 m | 8,000 ft 2,438 m |
|---|---|---|---|---|---|---|---|
| Add to chart psi | +0.5 | +1 | +1.5 | +2 | +2.5 | +2.9 | +3.8 |
Columns are volumes of CO₂. "—" means the beer already holds that much at zero gauge pressure. Above 20 psi (marked) it carbonates fine but pours badly on home-length lines. Standard regression of the ASBC solubility data; altitude from the standard atmosphere. Computed by this page's engine.
How long does it take to carbonate a keg?
The honest answer is that the chart doesn't know, and neither does anyone quoting it. Henry's law fixes where the beer ends up; how fast it gets there depends on the surface area between gas and beer, whether the keg is shaken, how full it is, and how cold. The ranking pages say "about a week," "5–7 days," "10–14 days" and "1–2 weeks" for the same setting, which is a tell: those are experiences, not calculations. At serving pressure the beer approaches the target from below and can't overshoot, so the range is harmless — pour a test glass after ten days and give it a few more if it's soft.
Burst carbing is the opposite case, and this is where the chart earns its keep even though it has no clock. "Set it to 30 psi for a day, then drop to serving pressure" works because a bigger pressure difference pushes gas in faster. But look at what 30 psi is heading for:
| Regulator at 38 °F | Equilibrium the beer is heading for | Versus a 2.5 target |
|---|---|---|
| 12 psi · 0.83 bar | 2.6 volumes | lands there and stops |
| 20 psi · 1.38 bar | 3.3 volumes | overshoots by 0.8 if left |
| 30 psi · 2.07 bar | 4.3 volumes | overshoots by 1.8 if left |
| 40 psi · 2.76 bar | 5.3 volumes | overshoots by 2.8 if left |
| 50 psi · 3.45 bar | 6.3 volumes | overshoots by 3.8 if left |
Beer at 38 °F, sea level. The equilibrium is where the beer would settle if the pressure were left on indefinitely. Computed by this page's engine.
At 38 °F, 30 psi is an equilibrium of 4.3 volumes — nearly double an ale's target; 40 psi is heading for 5.3. The method only works because you turn it off partway there, and where "partway" falls at hour 24 depends on the keg's headspace, how full it is, and whether anyone rolled it. The most careful burst-carb write-up we found (Brülosophy's) tabulates its own results as 30 psi for 16 hours, 40 for 12, 50 for 8 — and then says plainly there is "no good way to predict CO₂ volume accurately." That's correct, and it's why this calculator won't print an hour. Burst carb if you like; just know you're racing toward 4.3 volumes and the only brake is the clock.
Does altitude change the psi?
Yes, and the two calculators that mention it say "add 0.5 psi per 1,000 feet" without saying why — which makes it easy to dismiss as fussiness. Here's the why. A regulator gauge reads pressure above the air around it, and the beer doesn't care about the air around it; it responds to the absolute pressure of CO₂ on its surface. At sea level the air contributes 14.7 psi to that, and the chart's numbers silently include it. At 4,000 ft the air is only 12.7 psi, so a gauge reading of 11.2 delivers 2 psi less to the beer than the chart intended, and it carbonates short. The fix is to add back what the atmosphere took away:
| Elevation | Air pressure | Add to chart | "0.5 per 1,000 ft" rule | 2.5 vol at 38 °F |
|---|---|---|---|---|
| 0 ft · 0 m | 14.7 psi | +0 psi | +0.0 psi | 11.2 psi |
| 1,000 ft · 305 m | 14.2 psi | +0.5 psi | +0.5 psi | 11.8 psi |
| 2,000 ft · 610 m | 13.7 psi | +1 psi | +1.0 psi | 12.3 psi |
| 3,000 ft · 914 m | 13.2 psi | +1.5 psi | +1.5 psi | 12.8 psi |
| 4,000 ft · 1,219 m | 12.7 psi | +2 psi | +2.0 psi | 13.2 psi |
| 5,000 ft · 1,524 m | 12.2 psi | +2.5 psi | +2.5 psi | 13.7 psi |
| 6,000 ft · 1,829 m | 11.8 psi | +2.9 psi | +3.0 psi | 14.2 psi |
| 8,000 ft · 2,438 m | 10.9 psi | +3.8 psi | +4.0 psi | 15 psi |
International Standard Atmosphere; weather moves the real figure by a few tenths either way. Computed by this page's engine.
The Brewers Association's draught manual puts it the same way — "carbonation is proportional to absolute pressure, not gauge pressure" — and uses a rule of 1 psi per 2,000 ft, with the same 4,000-ft example this page reproduces: 11.3 psig at sea level becomes 13.3. The folk rule is a straight line through a curve, but the curve is gentle: it stays within a quarter psi of the real atmosphere all the way to 8,000 ft, reading slightly high. At a mile up, add 2.6 psi. At 1,000 ft, 0.5 — less than your gauge can show, which is why the calculator treats elevation as optional.
Volumes of CO₂ by style
| Style | Volumes | Regulator at 38 °F | In bar |
|---|---|---|---|
| British ales, cask-style | 1.5–2 | 0.7–6 psi | 0.05–0.41 |
| Porter, stout | 1.7–2.3 | 2.8–9.1 psi | 0.19–0.63 |
| Belgian ales | 1.9–2.4 | 4.9–10.2 psi | 0.34–0.7 |
| American ales | 2.2–2.7 | 8.1–13.3 psi | 0.56–0.92 |
| European lagers | 2.2–2.7 | 8.1–13.3 psi | 0.56–0.92 |
| German wheat beer | 3.3–4.5 | 19.6–31.9 psi — long lines or colder | 1.35–2.2 |
Ranges as published by Brewer's Friend and the BJCP; the same table the priming-sugar calculator uses, so a bottled and a kegged batch of the same beer are aimed at the same number. Sea level. Computed by this page's engine.
The wheat-beer row is the one to notice. Holding 4.5 volumes at 38 °F takes 32 psi, and a pressure that carbonates is also the pressure the beer is served at. Through a normal 5-foot picnic line that pours foam; balancing it takes a very long line, a colder fridge, or accepting the low end of the range. That trade-off is the beer-line-length calculator's job, and it's the next tool in this group.
Why is my keg flat — or all foam?
Flat after a week, in rough order of likelihood: the beer was warmer than the row you read (see the table above — the clock starts when it's cold); a leak, which a regulator can't tell you about because it just keeps feeding gas — spray the fittings with soapy water and watch for bubbles; not enough time, if the keg is full and undisturbed; and the gauge itself, which on a cheap regulator can be a psi or two off. If you're above 3,000 ft and never corrected for it, that's a half-volume on its own.
Foam: a burst-carb setting that stayed on too long (vent the keg, let it sit a day at serving pressure, vent again); a fridge colder than you thought, so the "right" pressure held more gas than you wanted; or — most often — a beer that is carbonated exactly right and simply served through too short a line. Turning the regulator down fixes the pour for a day and then flattens the beer, because the beer will out-gas toward whatever the new pressure holds. The pressure is the carbonation; the line is the pour. Fix the one that's wrong.
- Know your keg's rating and stay under it. Burst pressures in the 30s are within what a sound ball-lock keg is built for; a keg with a tired lid gasket or a corroded pressure-relief valve is not. Test the relief valve by hand before you crank it.
- Never shake a keg with the gas line connected unless there is a check valve in the line — beer can back up into the regulator. Pressurize, disconnect, then roll.
- CO₂ displaces air. A leaking cylinder in a closet-sized keezer room or a cellar is a real hazard, not a theoretical one. Tanks stay upright and secured; a leak you can hear is a room you leave.
- Every fitting gets the soapy-water test the first time it's under pressure. A 5-lb cylinder can empty overnight through a leak too small to hear.
The honest summary: the chart is exact and the world around it isn't. The beer isn't at the temperature you think for the first day, the fridge cycles, the gauge is approximate, and the air you're standing in changes what "zero" means. Set for the beer's real temperature at the warm end of its swing, correct for altitude if you have any, and give it the time nobody can compute — then trust the glass.