Do hot vintages really taste riper?

Everyone in wine says it. Hot years make jammy, baked, alcoholic wines; cool years make lean, tart, green ones. It sounds obvious. But I couldn't find anyone who had checked it against the actual weather, so I did.

The problem with the obvious answer

You can't just compare wines from hot places with wines from cool places. Napa tastes riper than the Mosel for a hundred reasons that have nothing to do with one summer. Different critics also have different vocabularies, Cabernet is described differently from Riesling, expensive wines get different words from cheap ones, and the whole industry drifted towards riper styles in the 2000s.

So the test has to compare a region with itself. Was Bordeaux described as riper in its hot years than in its cool years, by the same critics, for the same grapes, at the same prices, after allowing for the general style of that year everywhere? If the weather matters, that comparison should show it.

What I measured

• The reviews. 54,542 Wine Enthusiast tasting notes from a public dataset, covering vintages 1997 to 2016, written by 15 critics across 43 regions. I kept a region's vintage only if it had at least 10 reviews, which leaves 564 region-vintages.
• The weather. Growing-season temperature for each region and vintage from ERA5-Land, the Copernicus climate reanalysis: April to October in the north, October to April in the south. Each vintage is measured against that region's own 1994 to 2017 average, so “+1°C” means one degree warmer than normal for that place.
• The language. Two word lists. Ripe words like jammy, baked, raisin, ripe, hot and opulent; fresh words like crisp, tart, green, bright, zesty and taut. Each note scores its ripe words minus its fresh words.

The model then compares each region's vintages with one another while holding the critic, the grape, the price and the vintage year fixed.

The result

Hotter seasons push the language riper, clearly and consistently. For every extra 1°C, the words most tied to heat show up much more often, and the words tied to freshness show up less.

“Baked” is the strongest signal: +39% per degree, from a base of about 3 notes in 100. “Ripe” is far more common, in roughly one note in five, so its +8% is a smaller change on a much bigger number. On the other side, “tart” (−15%), “green” (−14%) and “crisp” (−10%) all fall away.

Some words don't move. “Fresh” and “bright” barely change, which makes sense: critics use them as general praise as much as a description of acidity.

Put together, a season 1°C warmer than normal adds about 12 extra ripe-over-fresh words for every 100 reviews of that region and vintage. That is not much for any single bottle. Across thousands of notes, it is unmistakable.

Is it real?

Two checks. First, the statistics: the main effect is about five times its standard error, whether or not I control for the general style of each year. Second, a harder test. I kept every region's real list of temperatures but shuffled which year got which weather, 2,000 times, and refitted each time. A random shuffle matched the real result in under 1 in 2,000 tries. If the weather had nothing to do with the words, that would be very unlikely.

The weather is moving

Everything above compares a vintage with its region's normal. The problem is that normal keeps changing. Across the 43 regions in this study, growing seasons warmed by about 0.48°C per decade between 1995 and 2025. The last ten seasons averaged 1.0°C warmer than the first ten, and the five warmest were all recent: 2022, 2020, 2017, 2018 and 2024.

This is not one or two regions. Every one of the 43 regions has a warming trend, and in 38 of them it is clearly larger than the year-to-year noise. Yakima Valley warmed fastest, at 0.88°C per decade. The slowest were cool coastal valleys like Santa Maria Valley (+0.09°C), where the change is small enough that it could still be chance.

Are we normalising baked wine?

Put the two findings together. On the trend, seasons now run about 0.9°C warmer than the 1997 to 2016 vintages the reviews come from. If critics keep reacting to heat the way they did, and the warming keeps its pace, the language moves with it.

On that path, “baked” would turn up in roughly 53 reviews in every 1,000 by 2050, against 29 in the period I studied: close to double. “Tart” would fall from 58 to about 40. “Jammy” points the same way, rising from 23 to about 28, but its link to temperature was too uncertain in the data to put much weight on.

There is a quieter effect the numbers cannot show. Critics describe a wine against what they are used to. If riper, more alcoholic wines become the everyday standard, “baked” stops being remarkable and may stop being written down at all. The words could drift less than the wine does. That is exactly why it is worth measuring the weather directly, rather than relying on how a vintage is talked about.

None of this is fixed. Growers are already picking earlier, moving to higher and cooler sites, managing canopies for shade and planting later-ripening grapes. The flattening in the very hottest vintages above may be a first sign of that. But the direction of the weather is clear, and it is pushing the style of wine one way.

What it doesn't show

• One publication, mostly New World. The biggest regions are Napa, Washington's Columbia Valley, Bordeaux, Mendoza and the Russian River Valley. It would be worth repeating with European critics.
• Word lists are blunt. “Hot” can mean heat from alcohol or a hot year; “green” can mean a herb or an unripe tannin. I also tried a more sophisticated language model that scores whole sentences. It picked up style, not ripeness, and found nothing. The simple word counts worked better.
• The very hottest years flatten out. The ripest-sounding vintages are around 0.5 to 0.75°C above normal; the hottest band dips back. That could be noise, or producers in hot years picking earlier to keep freshness.
• It measures how critics write, not what is in the bottle. If a critic expects a hot vintage to taste riper, some of this could be expectation. Blind tastings would help separate the two.
• The projection is a straight line. It assumes warming keeps its 1995 to 2025 pace and that critics react to +2.1°C the way they reacted to smaller swings. By 2050 that is at the edge of anything in the data, and winemakers will adapt in the meantime.

Conclusion: what counts as a hot year is changing

Two things are true at once. Critics really do taste the weather: a season 1°C warmer than normal brings measurably more “baked”, “hot” and “ripe”, and less “tart”, “green” and “crisp”. And the seasons themselves are warming, by about 0.48°C a decade across these regions, so that today's growing seasons already run around 0.9°C warmer than the vintages the reviews describe. What used to be a hot vintage is on its way to becoming an ordinary one.

That makes two questions worth asking of any region: how fast is it warming, and how much does it swing from one year to the next? A region that warms slowly and keeps its seasons steady is one where the style in the bottle is least likely to surprise you.

The steady ones. 7 regions are better than the median on both counts: Marlborough, North Fork of Long Island, Colchagua Valley, Finger Lakes, Anderson Valley, Sonoma Coast and Santa Maria Valley. Most of them have water close by, whether the cold Pacific along the California coast, the sea around Marlborough or the lakes and sound of New York State, which tends to take the edge off hot spells. Their vintages vary least, and they are warming more slowly than most.

Europe. No European region in this study clears both bars. The Alto Adige and Veneto have the steadiest seasons, but they are warming at or above the average. Bordeaux is the opposite case: warming more slowly than most (+0.43°C a decade) but swinging more from year to year.

Where the vintage matters most. Mendoza Province, Ribera del Duero, Beaujolais, Burgundy and Rhône Valley swing the most from one season to the next. In those regions the year on the label tells you more, and it is worth checking what the season did before you buy.

So, three practical suggestions:

• For dependable everyday buying, lean towards the steady regions above, where one vintage is most like the next.
• In the volatile regions, buy by vintage, not just by producer, and treat a very warm year with some caution if you prefer a fresher style.
• Inside warm regions, look for cool sites: altitude, ocean breezes or late-ripening spots. The Haute Vallée in Limoux is a good example: southern France, but about 1.5°C cooler than Serralunga d'Alba, and it shows in the freshness.

One caveat: these 43 regions are the ones the review data covers, so the list leans towards the Americas, and steady weather says nothing about quality on its own. It tells you how predictable a region is, not how good.

Why this site does what it does

This is the reason every note here carries the weather. If one summer can shift how a wine is described this much, the season is part of the story of the wine. So each note shows what the growing season actually did at that vineyard, measured the same way, and says plainly when the data does not support the obvious explanation.