What Is a Vintage Wine? What the Year Really Tells You
Updated
The year on a bottle is a weather report, not an age. What is a vintage wine? One made from grapes picked in a single named year, and the number records that harvest, not the bottling and not how long the bottle has sat anywhere.
The law puts a floor under that claim, and the two big rulebooks set it differently. In the United States, 27 CFR 4.27 requires at least 95 percent of the wine to come from grapes harvested in the labeled calendar year when the label carries a viticultural area such as Napa Valley, and at least 85 percent when the appellation is broader, such as a state or a county. In the European Union and the UK, Article 49 of Commission Delegated Regulation 2019/33 sets one line for everything: 85 percent. Blend across two or more years and the wine carries no vintage at all. So a 2024 bottled last spring is a vintage wine, and a thirty-year-old NV Champagne is not.
Does "vintage" mean old?
No. Neither rulebook mentions age, and a vintage date tells you nothing about whether a bottle is mature. The confusion is borrowed from other markets. Wiktionary lists the wine senses first: "the yield of grapes or wine from a vineyard or district during one season" and "the harvesting of a grape crop and the initial pressing of juice for winemaking". Then comes the adjective everyone else uses: "classic, or old enough to be recognizably outdated but not old enough to be antique", with the trade conventions attached. Cars qualify as vintage when "built between the years 1919 and (usually) 1930", clothing when "produced between the years of 1945 and 1970", watches when "produced between the years 1870 and 1980". Wine never moved with them. Wiktionary adds that the word comes through Old French vendage from the Latin vindēmia, a gathering of grapes.
What vintage wine means in law is a date of harvest, not a date of readiness. So how old is a vintage wine? Exactly as old as the number says, and not a day more. Whether it is ready to open is a drink-window question, and our drink-now list answers it bottle by bottle.
One clause in 27 CFR 4.27 is worth knowing because it stops the obvious cheat. "In no event may the quantity of wine removed from the producing winery, under labels bearing a vintage date, exceed the volume of vintage wine produced in that winery during the year indicated by the vintage date." A winery cannot make more 2019 than 2019 gave it.
What is a wine vintage actually telling you?
Weather, mostly, and the effect is large enough to trade on.
Orley Ashenfelter, David Ashmore and Robert Lalonde published the arithmetic in CHANCE in 1995: the quality of the vintage for red Bordeaux wines, as judged by the prices of mature wines, can be predicted by the weather during the growing season that produced the wines. Their regression runs on the vintages of 1952 to 1980, excluding 1954 and 1956, and uses four variables.
| Variable | Coefficient | What it says |
|---|---|---|
| Age of vintage | +0.0238 | older vintages fetch more |
| Average temperature over the growing season, April to September | +0.616 | the largest coefficient in the equation |
| Rain in September and August | -0.00386 | harvest rain is the reliable spoiler |
| Rain in the months preceding the vintage, October to March | +0.001173 | a wet winter helps |
Ashenfelter shows that age on its own explains 21.2% of the variation in price across vintages. Add the three weather terms and the R² reaches 0.828. The authors compress it into one sentence: Great vintages for Bordeaux wines correspond to the years in which August and September are dry, the growing season is warm, and the previous winter has been wet. The 1961 vintage, still the reference point for the region, had the lowest August and September rainfall in Bordeaux history.
Why do two vintages of the same wine cost different money?
Because the market prices the growing season twice: badly when the wine is young, properly when it is mature.
Ashenfelter states the gap bluntly in the same paper. "Bad" vintages are overpriced when they are young, and "good" vintages "be underpriced when they are In". The worked example is 1967. On the weather, those wines should have sold at about one half the price of an average of the 1961, 1962, 1964 and 1966 vintages. They entered the auction markets in 1972 at about 50% more than expected and slowly drifted down in relative price over the years. The conclusion the authors draw: Although the evidence suggests that the market for older wines is relatively efficient, it implies that the market for younger wines is very inefficient.
That is the entire case for reading a vintage before a campaign price rather than after, which is why en primeur mechanics deserve separate scrutiny.
The most famous reversal ran the other way. Robert Parker called the 1982 Bordeaux vintage superb, "contrary to the opinions of many other critics, such as San Francisco critic Robert Finigan, who felt it was too low-acid and ripe", and prices of 1982 Bordeaux "remain consistently higher than other vintages". Our Bordeaux 1982 page carries the realised results behind that sentence rather than the story.
Why do some serious wines carry no vintage at all?
Because consistency is the product. A non-vintage wine is "usually a blend from the produce of two or more years", and in Champagne that blend is the house style.
The Comité Champagne is clear that declaring a vintage is a choice, not a consequence. "Ce choix est celui de l'élaborateur et de l'élaborateur seul", the choice belongs to the producer and the producer alone, taken "lorsqu'une année mérite d'être magnifiée", when a year deserves to be magnified. The price of that choice is time. Every champagne must spend "at least 15 months maturing in the winemaker's cellars before release". A vintage cuvée must spend "un minimum de 36 mois en cave". In practice the Comité puts the working figures at "2-3 years for non-vintage wines and 4-10 years for vintage Champagne".
So a vintage bottling from a house like Dom Pérignon is not the same wine with a year added. It is a different production decision with several extra years of cellar cost sunk before anyone is paid.
One footnote most explainers skip: the 85% test ignores what goes in at the end. Article 49 requires that "at least 85 % of the grapes used to make those products have been harvested in the year in question", then excludes from that calculation any quantity used in sweetening, in "expedition liqueur" or in "tirage liqueur". The dosage does not have to come from the vintage.
What does a declared vintage mean in Port?
It means a house made a decision, then waited two winters to make it.
Port is the clearest case of a vintage date working as a quality claim rather than a calendar entry. Houses declare a vintage only in the years that earn it. Taylor's describes the mechanics: the wines "are then left to age for two winters in oak vats", "in their second Spring, they are tasted again", and "historically, Taylor's has only declared about three Vintage years per decade".
Three declarations a decade is a different signal from a vintage date on a claret, where the year appears whatever the weather did. Read the two the same way and you will overpay for one and overlook the other.
Does a warm year still mean a good year?
Less reliably than it did, and the longest record in existence says so.
Thomas Labbé and colleagues rebuilt the Beaune grape harvest series from 1354 to 2018 and published it in Climate of the Past in 2019. Over the first six centuries of the record, "grapes were on average picked on 28 September from 1354 to 1987". Then the series breaks: harvest dates "were 13 d earlier (15 September) during the last 31-year-long period of rapid warming from 1988 to 2018". The series tracks April to July mean temperatures in Paris at r = -0.76 over 1659 to 2018, and the authors' summary is the line to keep: "warm extremes in the past were outliers, while they have become the norm in the present time".
Now set that against the Bordeaux equation, fitted on 1952 to 1980, whose largest coefficient is growing-season warmth. Writing in 1995, those authors had already spotted the drift: of the vintages from 1981 to 1992, all but 1986 came from a growing season warmer than what is historically normal.
A model whose largest term is heat loses its power to separate vintages once every vintage is hot. Anyone still buying off a vintage chart calibrated on the old distribution is using a ruler that has stopped measuring.
North and south do not share a vintage year
The evidence sits in public reporting, and the gap is a whole growing season.
Wine Australia reports a 2026 Australian winegrape crush of 1.27 million tonnes, in the national vintage report it published on 15 July 2026. On that date the fruit for a 2026 Burgundy was still on the vine: the modern average picking date in the Beaune series is 15 September. A 2026 from Australia was picked, pressed and counted before a 2026 Burgundy was cut from the vine. Same number on two labels, two different growing seasons.
The threshold does not move with the hemisphere. Australia and New Zealand sit on the same 85% line as the EU. The law notices one edge case: where a product is traditionally made from grapes harvested in January or February, Article 49 puts the previous calendar year on the label. Everything else takes the calendar year of the pick.
How do you read a vintage chart without being misled?
As one variable of two, never as the answer.
Wine Spectator describes what its charts are doing: they "summarize the quality and character of wines from a particular region for each individual vintage", and they are "by necessity, general in nature". That is a fair description of a regional average, and a regional average is half the information.
The other half is the producer. Ashenfelter found that knowledge of the chateau and vintage provides most of the information about the quality of the wine, and that the explained variance from a regression of log price on chateau and vintage dummies is over 90%. Two variables, both load-bearing. A chart hands you one of them. The Burgundy producer atlas is where the other one lives, and the market index is where the two get compared across regions.
When does a vintage tell you to drink it?
It starts the clock and nothing more. Everything after the harvest is storage, closure and structure, which is why one year can be open in one cellar and shut in another.
Fill level is the visible half of that history, so read the ullage bands on older bottles before you treat a vintage as a guide to readiness.
The CHANCE paper contains one number every collector should own. Across chateaux with at least 20 vintages in the sample, Ashenfelter reports a real rate of return on holding Bordeaux of between 2% and 4% a year, and the reciprocal of that return is the size of a self-funding cellar: about 28 bottles in a perpetual cellar that was intended to support the consumption of one bottle per year. Buying for aging on that arithmetic turns a cellar into a rotation instead of a hoard.
Open the right year, not the famous one
Knowing which year to open is worth more than knowing which year scored highest, and it is the part no chart gives you.
Every wine here carries a vintage maturity chart: where each year sits in its drink window, how the market has priced it against the years either side, and where the crowd is drinking early or holding too long.
Our newsletter sends the vintages that moved into their window that month, with the price history beside them, so you open bottles at their peak and buy the years the charts have not caught up with. Start with Bordeaux 1982 for the full treatment on a famous year, or with the drink-now list if the bottles are already downstairs and the only question left is when.