Buying coffee in small amounts costs more per kilogram and forces you to think about it every fortnight. Buying a large bag costs less and guarantees that the last third of it will not taste like the first third. That is the real decision hiding underneath the question of when coffee goes stale, and it has no single answer, because the clock is set by the format the coffee is sitting in rather than by the coffee itself.
There is no one date. Whole beans in a sealed valve bag are at their best somewhere between three and twenty-one days from the roast date and are noticeably declining by four to six weeks. Once you open that bag, plan to finish it inside two to three weeks. Ground coffee starts losing aroma the moment it leaves the grinder and is badly diminished within a day, which is why pre-ground retail coffee is compromised before it reaches the shelf. Green unroasted coffee holds six to twelve months from harvest. Brewed coffee starts changing the moment it is poured. The best-before date printed on a supermarket bag is not answering this question and was never trying to.
The table
| Format | Clock starts at | Good while | What fails first |
|---|---|---|---|
| Green, unroasted | Harvest | 6 to 12 months | Acidity fades, papery and baggy notes arrive |
| Whole bean, sealed valve bag | Roast date | Best 3 to 21 days, meaningfully declining by 4 to 6 weeks | The lightest aromatics |
| Whole bean, bag opened daily | First opening | 2 to 3 weeks | The same, faster, because you refill the bag with air each time |
| Whole bean, frozen in single-use doses | Freezing | Months | Very little, provided you never refreeze |
| Ground at home | The grinder | Loss starts immediately, severe within a day | Fragrance, then everything behind it |
| Pre-ground retail, sealed | Packing | Months of shelf life, none of it good | It was compromised at the factory |
| Sealed single-dose capsule | Packing | Months | Holding flavour is the format's one real achievement |
| Instant | Packing | A very long time. It is a dried solid | It never carried those compounds to begin with |
| Brewed, in a preheated vacuum flask | The brew | Longer than a jug, and we will not give you a number we have not measured | Aroma, then balance |
| Brewed, on a hot plate | The brew | Minutes | It keeps cooking, and cooked coffee does not come back |
| Espresso in the cup | The shot | About four minutes | Crema first, then the balance of the drink |
| Cold brew concentrate | Filtering | A food-safety question, not a flavour one. See below | Not our call to make |
Why the numbers are shaped like that
Three separate mechanisms are running at once, and most freshness advice collapses them into one word.
The first is loss. A green bean carries something on the order of two hundred volatile compounds. Roasting leaves it with more than a thousand. Volatile means exactly what it says: those molecules are inclined to leave. Jessica Easto's work on tasting supplies the detail that makes staling predictable rather than mysterious. Weight decides the order of departure. The lighter a molecule is, the sooner it goes, and the lightest ones happen to carry the top of a coffee's register, its floral and honeyed side. So a coffee does not fade evenly. It loses its top end while the body and the bitterness stay put, which is why a stale light-roast Ethiopian collapses into almost nothing while a stale dark roast just reads as flat and heavy. The bright coffee had further to fall.
The second is gain, and it is the unpleasant one. Oxygen reacts with the oils in the bean and produces compounds that were not there before. Flat and papery, closer to a damp box than to coffee. Stale coffee is a different substance from fresh coffee, not a quieter version of it.
The third is the reason none of this happens as fast as it should. Kenneth Davids, writing about home roasting, treats the carbon dioxide still inside a roasted bean as a temporary defence. The gas is on its way out, and for as long as it is leaving it holds the interior against the oxygen that would otherwise be filling that space. The bean shields itself with something roasting made. The corollary is uncomfortable and worth stating. Degassing and staling run on the same clock. A bean that has stopped venting has also stopped defending itself, and by then most of what you paid for has already left.
Grinding ends all three arguments at once. It multiplies the exposed surface area by something on the order of a thousand times, releases the remaining gas quickly, and opens the interior to air. The familiar rule is to brew within fifteen minutes of grinding, and the honest thing to say about that number is that it is trade convention rather than a measured threshold. We could not find a measurement study behind it and we are not going to dress one up. What is well founded is the instruction underneath it, which is Davids's single most useful line in a whole book about roasting: grind nothing until the minute you are ready to brew.
The two edges of the window
Freshness is not a deadline with one side. There is a floor as well as a ceiling, and the floor is the part people get wrong.
Coffee straight out of the roaster is still venting gas, and that gas obstructs extraction. Espresso misbehaves worst, gushing and refusing to hold a stable shot. Moldvaer's rest guidance after a home roast is one to two days for filter and about a week for espresso, and we have argued at more length that most people open the bag too early rather than too late.
Then there is a genuine disagreement, and it is worth naming rather than smoothing over. Davids, writing for home roasters, puts the peak at roughly twenty-four hours out of the roaster and sees nothing but decline afterwards. Most current specialty practice puts the peak later, from around three days out to three weeks. Both camps are describing something real. Davids is writing about drip and filter, where the venting gas causes less trouble, and about coffee roasted in your own kitchen, where day-one aroma is the whole reason for the exercise. The specialty window is set largely by espresso, where day-one coffee is unusable. If you brew filter, lean early. If you pull shots, lean late.
Whichever edge you sit on, you can check without equipment. Wet the grounds and watch. Fresh coffee foams and rises as the trapped gas escapes, and a bed that stays flat and resists wetting has nothing left to give. That is a diagnostic rather than a technique, and it is the cheapest freshness test there is.
What the best-before date is actually telling you
Nothing about flavour. Supermarket coffee routinely carries a best-before date twelve to twenty-four months out because roasted coffee is shelf-stable, and shelf-stable is a statement about storage rather than about quality. A date that far out is a confession about what the producer expects you to notice.
The useful number is the roast date, and the useful object is the bag. A one-way valve on a foil bag lets carbon dioxide out without letting oxygen in, which tells you the coffee was packed while still degassing and therefore packed recently. A hard vacuum brick with no valve tells you the opposite: the coffee had finished venting before it was sealed, which means it sat somewhere first. Those two are alternative solutions to the same problem, and a bag claiming both is claiming something incoherent. Reading the valve and the roast date takes five seconds and outperforms most equipment upgrades.
Green coffee, and the claim we will not repeat
Unroasted coffee keeps far longer than roasted, which has led to a persistent claim in older books that it keeps indefinitely without harm. It does not. Specialty green is at its best within roughly six to twelve months of harvest, after which it slides into what the trade calls past crop: flattened acidity and a woody, papery overlay. Stored well it stays usable for a good deal longer. Usable and good are different words, and green coffee not keeping forever is one of the few places where the reference literature is simply wrong rather than dated.
What this article will not do
It will not tell you whether an old bag is safe to drink. Stale and spoiled are different conditions, and which one you have turns on storage we cannot see from here. Anything that has been kept in humidity, or that smells musty rather than merely dull, is outside the scope of a flavour article and outside what we are qualified to judge.
It will not give you a shelf life for cold brew. Cold brew is never heated, and its acidity is close to hot coffee's rather than lower: Rao and Fuller measured both methods between pH 4.85 and 5.13 in a laboratory study published in Scientific Reports in 2018. Neither of those two facts is a shelf life, and neither is reassuring. How long a jug of it keeps is a food-safety question, we have not verified current official guidance on it, and we are not going to print a number we cannot stand behind. Anyone telling you confidently that it keeps for two weeks should be asked where they got that.
And it will not pretend the freezer is settled. Freezing whole beans in sealed single-use portions is well supported for long storage, and the failure mode everyone attributes to freezing is actually repeated thawing, which condenses water onto the beans. Even here the sources disagree on the details, including whether to grind straight from frozen or thaw first, and Davids himself takes the more conservative side. We have set out the version we think holds up and the reasons for it, and we have flagged where the evidence is thinner than the confidence around it.
The honest recommendation
Buy the amount you will drink in two weeks, and buy it with a roast date on it. That single habit does more for the cup than any storage container, because no vessel slows the clock as much as simply not owning coffee long enough for the clock to matter. If you cannot buy that often, buy larger and freeze the excess in portions on the day it arrives, rather than after you have watched it fade for a month.