Something is wrong with the beer and you want to know what and why. The charts will tell you that butter is diacetyl and green apple is acetaldehyde, which is a start — but they list a dozen faults in one column as though they were comparable, and they are not. Three of them have published detection thresholds, and those three sit 333 times apart. That is the single most useful thing to know before you read any off-flavor table, including this one.
Start with when it appeared, not what it tastes like
Descriptors are ambiguous — "cidery" is acetaldehyde to one taster and a mild infection to another — but the timeline usually is not. A fault present in the first glass and never worse is a process fault, decided on brew day or during fermentation. A fault that grows over weeks in the bottle is oxidation or an infection, and those two are the only ones that get worse in storage. If every bottle from the batch tastes the same, suspect process; if they diverge, suspect packaging.
The off-flavor table
| Fault | Tastes like | Detected at | Where it comes from | What to do |
|---|---|---|---|---|
| DMS | Cooked or creamed corn, cooked vegetables | 30–50 µg/L | SMM from pale malt converts to DMS in the heat and stays if it cannot evaporate — a lidded kettle, a lazy boil, or slow chilling. | Boil hard and uncovered for 60 minutes, then chill fast. |
| Diacetyl | Butter, butterscotch, toffee; slick mouthfeel | 50–150 µg/L | Yeast makes it early and reabsorbs it late. Cold-crashing or racking too soon strands it. Also produced by Lactobacillus and Pediococcus infection. | A diacetyl rest — warm to 18–20 °C for two to three days before crashing. |
| Acetaldehyde | Green apple, cut pumpkin, cidery | 10,000–25,000 µg/L | An intermediate on the way to alcohol. Healthy yeast clears it; packaging a young beer locks it in. | Time on the yeast. Confirm a stable gravity before packaging. |
| Oxidation | Cardboard, papery, stale; sherry-like in strong beer | not published | Oxygen picked up after fermentation — splashing at transfer, headspace in a keg or bottle, a leaking seal. | Nothing, once it is there. Transfer gently, purge the receiving vessel, fill from the bottom. |
| Fusel alcohols | Solventy, hot, harsh; nail-polish remover | not published | Fermentation too warm, especially in the first 48–72 hours, and worse in strong beers. | Hold the low end of the yeast’s range while it is most active. Time softens it a little; nothing removes it. |
| Chlorophenols | Medicinal, plastic, band-aid, TCP | not published | Chlorine or chloramine in the brewing water reacting with malt phenols. Also wild yeast. | Treat the water — a crushed Campden tablet per 20 L removes both. Never rinse with untreated tap water. |
| Astringency | Harsh, puckering, dry, over-steeped tea | not published | Tannins from grain husks — sparge water too hot, mash pH too high, or squeezing the bag hard. | Keep sparge water under about 77 °C and mash pH in the 5.2–5.6 band. |
| Lightstruck | Skunk, struck match, cat musk | not published | Light reacting with hop compounds. Minutes in sunlight through clear or green glass is enough. | Brown bottles, and keep them in the dark. |
| Autolysis | Meaty, rubbery, yeast-extract | not published | Yeast breaking down after a long warm rest on the cake. | Package within a few weeks, or crash and rack off the yeast. |
| Sourness | Sharp acidity, vinegar, lactic tang | not published | Bacterial infection, usually from a scratched bucket, a tap, or a hose that was not cleaned before it was sanitized. | Nothing. Replace scratched plastic, clean before sanitizing, and disassemble taps. |
Thresholds from Brewfather's quality-control documentation, the only reachable homebrew source that publishes them with units; diacetyl's range contains Brew Your Own's 0.1 ppm. All three are shown in µg/L — parts per billion — so they can be compared on one scale, though acetaldehyde is more usually quoted as its equivalent 10–25 mg/L and diacetyl as 0.05–0.15. "Not published" means exactly that: no number is guessed here.
Why the thresholds matter more than the descriptors
Look at the three that have numbers. DMS is detectable at 30–50 µg/L. Acetaldehyde takes 10,000–25,000 µg/L — 333 times as much of it before anyone notices. Diacetyl sits between them at 50–150 µg/L.
That spread explains things the charts leave mysterious. It is why DMS is the fault people find in commercial lagers — a quantity of it too small to matter for any other compound is already over the line. It is why acetaldehyde tends to be all-or-nothing: you are either well under 10 mg/L and taste nothing, or you packaged a green beer and it is obvious. And it is why "I think I can detect a hint of X" is worth treating skeptically for anything but DMS. These are not one kind of problem and they do not deserve equal worry.
Butter or butterscotch: diacetyl, and when to start the rest
Yeast produces diacetyl early and reabsorbs it late, so almost every case in homebrew is a beer that was chilled or racked before the yeast finished tidying up. The fix is a diacetyl rest: warm the fermenter to 18–20 °C for two or three days. The part that gets left out is when — Brew Your Own says to start it when gravity is "two to five specific gravity points away from the target terminal gravity", which is a rule about your beer, not a date. So here is that rule as a reading:
| Your beer | Should finish at | Start the rest between |
|---|---|---|
| 1.040 · 75% yeast | 1.010 | 1.015 and 1.012 |
| 1.050 · 75% yeast | 1.012 | 1.017 and 1.014 |
| 1.060 · 75% yeast | 1.015 | 1.020 and 1.017 |
| 1.050 · 80% yeast | 1.010 | 1.015 and 1.012 |
| 1.070 · 72% yeast | 1.020 | 1.025 and 1.022 |
Terminal gravity from this site's engine for each OG and the strain's apparent attenuation; the window is BYO's two-to-five points above it. Use your own strain's published range in the attenuation calculator.
For the common case — a 1.050 ale on a 75% strain — terminal is 1.012, so you want the temperature up when the hydrometer reads about 1.017 to 1.014, while there is still active yeast to do the work. Warming after it has finished and settled works far less well, which is the usual reason a rest disappoints.
A rest fixes yeast-derived diacetyl. If the butter appears weeks later, or keeps intensifying, that is Pediococcus or Lactobacillus making it — an infection, not a fermentation mistake, and no amount of warmth will clear it. Slick mouthfeel with the butter points the same way. The fix is the cold-side cleaning regime, not the fermenter.
Green apple and cooked corn: the two that answer to patience and heat
Acetaldehyde is a normal step on the way to alcohol, and healthy yeast eats it. Green apple in a young beer is not a fault so much as an unfinished one — leave it on the yeast another week and take two readings three days apart before you package. If it is still there in a beer that has genuinely finished, look at pitch rate and oxygen, and at whether the batch stalled: our stuck fermentation guide covers that decision.
DMS is different: it is created by heat and removed by evaporation, so it is decided entirely in the kettle. The precursor keeps converting the whole time the wort is hot, which means a lid on the kettle, a simmer rather than a boil, or a slow chill all leave it in. A hard, uncovered 60-minute boil and a fast chill is the whole answer — and because the threshold is so low, this is the one fault where "close enough" is not. Pale lager malts carry more of the precursor than any other, which is why the flaw shows up in the styles least able to hide it.
Cardboard and sherry: oxidation, and what a purge is worth
Oxidation is the only fault on the table with no cure at all. Once oxygen has got into finished beer, the reaction is done and the flavor is coming — faster if the beer is warm, and unmistakable in a few weeks. Everything is prevention: transfer gently and without splashing, fill from the bottom, and push the air out of whatever the beer is going into.
For a keg, "push the air out" has a number. Filling with CO₂ and venting is a dilution, so each cycle leaves a fraction of what was there:
| Purge cycles | At 15 psi | At 30 psi |
|---|---|---|
| 1 | 49.5% of the air left | 32.9% of the air left |
| 2 | 24.5% of the air left | 10.8% of the air left |
| 3 | 12.1% of the air left | 3.6% of the air left |
| 4 | 6.0% of the air left | 1.2% of the air left |
| 5 | 3.0% of the air left | 0.4% of the air left |
Residual headspace air after repeated pressurize-and-vent cycles, from this site's engine. The CO₂ calculator costs the same cycles in grams of gas.
Two things fall out of that. Pressure does more per cycle than repetition does — two purges at 30 psi leave 10.8%, beating three at 15 psi (12.1%) while using fewer fills. And the returns stop mattering: past four or five cycles you are chasing tenths of a percent, and the oxygen that will actually spoil the beer is the dissolved oxygen you splashed in at transfer, which no amount of headspace purging removes. If you bottle, the same logic applies with worse tools — fill quietly through a wand, leave a consistent small headspace, and accept that bottled beer ages faster than kegged.
Some of these are the style, not a fault
Worth saying, because the charts do not. The American Homebrewers Association's own guidance is that a low level of several of these is on-style rather than wrong:
- DMS: Low levels are on-style in American and international lagers, some helles and pilsners.
- Diacetyl: Small amounts are on-style in Czech lagers, bitters, some stouts and Scottish ales.
- Acetaldehyde: Low apple character is on-style in American light lager and kellerbier.
A Czech lager without a whisper of diacetyl is arguably missing something, and a helles with a trace of DMS is a helles. The threshold is where a compound becomes detectable, not where the beer becomes bad. Chase the faults that are genuinely spoiling your beer — and the ones on this page that have no cure, because those are the ones only prevention will fix.
The habits that prevent most of them
Nearly every fault above is decided by four things, and none is a purchase. Treat your water — one crushed Campden tablet per 20 L removes chlorine and chloramine, and takes the plastic and band-aid flavors off the table permanently. Pitch enough healthy yeast and hold the temperature, especially through the first 72 hours, which is where fusels and most diacetyl are decided. Boil hard and uncovered, then chill fast. And clean before you sanitize — sanitizer does nothing to a surface with a film on it, which is why scratched buckets and unloved taps are where sourness and lasting diacetyl come from.