How High Does a Ski Resort Need to Be? What 15 Winters Say About Altitude
Published 28 Aug 2026
“Book high” is the most repeated piece of ski advice there is, and on 15 winters of history across the 32 resorts we track it is true in a narrower way than it sounds. Mid-mountain altitude correlates with our season score at r = 0.67, which is a real relationship and also means altitude leaves more than half the difference between resorts unexplained. Strip out the glaciers, whose permanent ice is doing something altitude alone does not describe, and it falls to r = 0.56.
The more useful finding is that the correlation is not spread evenly. Altitude does almost all of its work in one narrow window, and outside that window it tells you close to nothing. Above about 2,100 m of mid-mountain, 15 of the 18 resorts here clear 186 of a possible 210. Below about 1,800 m, every resort lands between 137 and 155 regardless of how much higher one is than another. The entire argument about altitude lives in the 300 m in between, which is exactly where the largest number of Alpine resorts sit.
What does the score mean here?
It is the same season score as the snow-sure ranking, unchanged, so the two guides can be read against each other. Each of the 14 half-months from November to May contributes the count of winters, out of the last 15, in which the resort had a 30 cm or deeper base at its mid-mountain elevation by that date. A resort that cleared 30 cm in every half-month of every winter scores 210.
Nothing new is measured in this guide. What is new is the sample: that earlier ranking covered 22 Austrian and Italian resorts and deliberately said nothing about Switzerland, because the Swiss history had not been computed when it was written. It has been since, and this guide runs over all 32 resorts in all three countries.
Where does altitude actually buy anything?
Grouping the 32 resorts into bands of mid-mountain elevation shows the shape better than any single correlation does. The spread column is the one that matters: it is the gap between the best and worst resort inside each band.
| Mid-mountain | Resorts | Lowest score | Median | Highest score | Spread |
|---|---|---|---|---|---|
| Below 1,800 m | 6 | 137 | 139 | 155 | 18 |
| 1,800 to 2,099 m | 8 | 114 | 182 | 198 | 84 |
| 2,100 to 2,399 m | 9 | 186 | 191 | 200 | 14 |
| 2,400 m and above | 9 | 139 | 199 | 210 | 71 |
Two bands are tight and two are wide, and they are wide for opposite reasons.
Below 1,800 m the spread is 18 points across six resorts. Zell am See at 1,350 m and Mayrhofen at 1,565 m are 215 m apart and 17 points apart, which is close to nothing. Down here altitude has already stopped being the variable, because none of these resorts is high enough for it to be decisive and all of them depend on the same thing: whether the winter turned up.
Between 2,100 and 2,399 m the spread is 14 points across nine resorts, every one of them between 186 and 200. This is the band where the question answers itself. Whatever else separates Ischgl, Livigno, Laax and Davos Klosters, natural snow reliability does not.
The 1,800 to 2,099 m band is where the choice is real, at a spread of 84 points, and that is the expected result. The genuinely surprising one is the top band. Going above 2,400 m does not tighten the distribution further, it widens it again, from a spread of 14 in the band below to 71. Its nine scores run 139, 169, 182, 192, 199, 199, 202, 206, 210. Those bottom three are Cortina d’Ampezzo at 2,470 m, Sestriere at 2,400 m and Kaprun at 2,450 m, and the first two are the reason the band misbehaves: above 2,400 m you start picking up resorts whose altitude is compensating for a dry position rather than compounding a wet one. Height that exists because the snow line is high is not the same asset as height that sits under a storm track.
The test that takes altitude out of the question
Six resorts on this site have a mid-mountain elevation of exactly 2,000 m. Because our history is modelled at that elevation, all six were queried with the same altitude, which makes them as close to a controlled comparison as an observational dataset gets. If altitude were the dominant term, these six would score alike.
| Resort | Country | Season score | Snowfall days, Nov to Dec | Base by 31 Dec | Median depth, 31 Dec |
|---|---|---|---|---|---|
| Bormio | Italy | 193 | 17.3 | 15 of 15 | 118 cm |
| Serfaus-Fiss-Ladis | Austria | 191 | 22.3 | 15 of 15 | 92 cm |
| Lech-Zürs | Austria | 173 | 23.6 | 12 of 15 | 81 cm |
| Obertauern | Austria | 155 | 22.2 | 13 of 15 | 61 cm |
| Madonna di Campiglio | Italy | 135 | 14.0 | 10 of 15 | 53 cm |
| Val Gardena | Italy | 114 | 11.6 | 8 of 15 | 40 cm |
The spread is 79 points out of 210, at one identical altitude. On the most expensive date in the calendar, the top of this list is a certainty and the bottom is a coin flip, and no part of that difference is height.
The mechanism is visible in the snowfall column, but not in a straight line, and the exception is worth stating rather than smoothing over. Val Gardena and Madonna di Campiglio get the fewest snowfall days of the six, and they finish last, which is the Dolomite precipitation shadow doing exactly what you would expect. Yet Lech-Zürs records the most snowfall days of any of the six and still finishes third, well behind Bormio, which sees six fewer days a season. Frequency is not amount, and it is not retention either. Bormio’s Alta Valtellina position gives it fewer, colder, better-banked storms; the Arlberg gives Lech more frequent ones that also spend more of the early season near the melting point at 2,000 m.
Which resorts beat resorts far above them?
Once the bands are set out, the individual reversals stop being anecdotes and start being the rule at work.
- Courmayeur at 1,956 m scores 198. Sestriere at 2,400 m scores 169. Sestriere is 444 m higher and 29 points worse. Both are recent additions to this site, and they sit either side of the same lesson: Courmayeur is at the mouth of the Mont Blanc massif, where systems stall, and Sestriere is on a pass at the dry western end of the Piedmont Alps, where its height is compensating for a thinner supply rather than adding to a generous one.
- Five resorts between 1,350 m and 1,565 m out-score Val Gardena at 2,000 m. Zell am See, Kitzbühel, Flachau, Saalbach-Hinterglemm and Mayrhofen are all 435 m or more below it and all score higher. This is the single most counter-intuitive row in the dataset, and it survives every way we have looked at it.
- Cortina at 2,470 m is beaten by both Lech-Zürs and Obertauern at 2,000 m. The second-highest non-glacier mid-mountain on the site sits 26th of 32.
None of these is a case of a low resort being reliable in absolute terms. Zell am See’s 138 is a weak score. It is simply a less weak score than Val Gardena’s 114, from 650 m lower down, which is the point.
Is this circular, given the model is asked for a specific altitude?
It would be, and this is the caveat that matters most in this guide, so it goes here rather than at the end.
Our 15-winter history comes from the Open-Meteo reanalysis archive queried at each resort’s mid-mountain elevation. Elevation is therefore an input to the number, not an independent observation of it. A guide that simply reported “higher resorts have more snow in our data” would be reporting its own assumption back to itself.
That is why the 2,000 m cohort is the load-bearing part of this piece and the band table is only the frame. Within that cohort the altitude input is held constant at a single value, so every point of the 79-point spread is the model responding to where in the Alps the resort is: how much precipitation arrives, at what temperature, and how it is held. The same applies to the reversals, which are cases where geography beats a 400 m to 650 m altitude advantage rather than merely offsetting a small one.
Two further limits are worth carrying. The score describes natural snow only, and snowmaking is not modelled, which flatters nobody and understates the resorts at the bottom of the table most, because those are the ones that invest hardest in it. The Dolomite areas in particular ski far better in December than their natural-snow rows imply. And a point value is not a mountain: aspect, shading and which side of a ridge a piste sits on move real depth by more than the gap between neighbouring rows here.
Does altitude matter more in some months than others?
It matters at the edges of the season and barely at all in the middle, which is the same shape the snow-sure ranking found and it holds with Switzerland added.
In the first half of February, 24 of the 32 resorts cleared 30 cm in all 15 winters and the worst on the site managed 14 of 15. There is no altitude question in February; there is barely a reliability question at all. In the first half of December only 12 of 32 are at 15 of 15, and the range runs down to 8 of 15. December is where a 2,000 m decision is made or lost.
The median depth by band on 15 December shows the same thing as a quantity rather than a probability: 45 cm below 1,800 m, 65 cm from 1,800 to 2,099 m, 79 cm from 2,100 to 2,399 m and 94 cm above that. The steps between bands are real, but so is the range inside the 1,800 to 2,099 m band, which runs from 34 cm to 98 cm.
So how high is high enough?
On this data, three numbers are worth carrying into a booking.
- 2,100 m of mid-mountain is where the argument mostly ends. Above it, 15 of 18 resorts here score 186 or better, and in the 2,100 to 2,399 m band specifically it is 9 of 9 with a spread of 14 points. The word “mostly” is doing real work: the three exceptions are all high resorts in dry positions, so above 2,100 m altitude stops being a worry without ever becoming a guarantee.
- 1,800 m is where altitude stops being the lever. Below it, everything on this site lands within 18 points of everything else. Choosing a 1,550 m resort over a 1,400 m one on snow-reliability grounds is optimising something that is not there. Choose on terrain, price and access instead, and check the specific half-month you are skiing.
- Between them, the massif decides, not the metres. In the 1,800 to 2,099 m band the spread is 84 points, wider than the gap between that band’s median and either neighbour. This is where you have to look past the elevation figure at whether the resort sits where the snow actually arrives.
And the general form of the rule, which the reversals above are all instances of: altitude preserves snow, and it cannot create snow that never fell. A high resort in a dry corner is keeping a small amount of snow well. A moderate resort under a wet airstream is keeping a large amount of snow adequately. Over 15 winters the second has usually been the better booking, and the first is the one the standard advice keeps recommending.
One thing this site cannot yet tell you honestly is whether a country-level version of any of this holds. Our Swiss resorts have a median season score of 199 against 177.5 for our Austrian ones, but they also have a median mid-mountain 228 m higher, and we built the biggest and highest Swiss resorts first. That is a sampling artefact, not a finding, and it will stay one until the coverage evens out.