There Is No Best Month to Ski the Andes. Valle Nevado's July Has Delivered Anything From 1 cm to 142 cm.

“What is the best month to ski Chile?” has a standard answer — July or August, sometimes June — and it is given with more confidence than the snow supports.

We took five of the best-known ski areas in the Andes and counted what fell in every month of the last twelve southern winters, one winter at a time. The honest answer is that there is no best month. There is a best month on average, and it is a poor guide to the month you will actually get.

The biggest month keeps moving

How often each month was the snowiest of the winter, across twelve winters:

	June	July	August	September
Valle Nevado 🇨🇱	5	2	4	1
Portillo 🇨🇱	6	1	4	1
Las Leñas 🇦🇷	5	2	4	1
Nevados de Chillán 🇨🇱	5	3	3	1
Cerro Catedral 🇦🇷	2	6	3	1

At Valle Nevado June won five times, August four, July twice and September once. No month wins even half the time anywhere on this list.

The twelve-winter average does put June ahead at Valle Nevado — 91 cm, against 55 cm for July. But that average is carried by a few enormous Junes: 227 cm in 2024 and 164 cm in 2020. Take the middle winter instead of the mean and June’s lead shrinks to 82 cm against 52.

Every month has failed

The same months, measured by how badly they can go:

July, last 12 winters	Worst	Typical	Best	Under 20 cm
Valle Nevado	1 cm	52 cm	142 cm	4 winters
Portillo	1 cm	50 cm	173 cm	3 winters
Las Leñas	9 cm	80 cm	198 cm	2 winters
Nevados de Chillán	6 cm	157 cm	406 cm	1 winter
Cerro Catedral	31 cm	100 cm	231 cm	none

In 2024, Valle Nevado took 227 cm in June and 1 cm in July. A skier who booked the second week of July that year on the strength of “July is peak season” arrived after a month with almost no new snow.

At Portillo, every month of the season has had at least one winter with less than 20 cm — June once, July, August and September three times each.

Set that against Japan’s January

January, last 12 winters	Worst	Typical	Best	Under 20 cm
Nozawa Onsen 🇯🇵	95 cm	166 cm	259 cm	none
Hakuba Happo-One 🇯🇵	97 cm	146 cm	172 cm	none
Niseko Grand Hirafu 🇯🇵	48 cm	78 cm	138 cm	none
St. Anton 🇦🇹	50 cm	86 cm	197 cm	none

Hakuba’s worst January of the last twelve delivered 97 cm — more than Valle Nevado’s typical June, its best month on average. That is the real difference between the two hemispheres, and a season total does not show it.

It shows up in whole winters as well. The gap between the best and worst of the twelve seasons:

	Worst winter	Best winter	Best ÷ worst
Nevados de Chillán	188 cm	1,081 cm	5.8×
Las Leñas	196 cm	1,034 cm	5.3×
Portillo	123 cm	584 cm	4.7×
Valle Nevado	112 cm	397 cm	3.5×
Cerro Catedral	160 cm	510 cm	3.2×
Zermatt	197 cm	513 cm	2.6×
St. Anton	204 cm	535 cm	2.6×
Niseko	166 cm	384 cm	2.3×
Nozawa Onsen	301 cm	635 cm	2.1×
Hakuba Happo-One	301 cm	609 cm	2.0×

Counted over twelve winters, the typical Argentine and Chilean ski area swings about one and a half times as much as Japan’s — a best-to-worst ratio of about 3.5 against 2.3.

Why the Andes are like this

Japan’s snow comes from a machine that runs most nights: cold air crossing the Sea of Japan and unloading on the mountains, week after week through January. A month there is made of dozens of small deliveries, and dozens of anything average out.

The central Andes get their winter from a small number of large Pacific storms. A month there is made of two or three deliveries, sometimes none. Whether a storm lands on the 30th of June or the 1st of July moves an entire month’s total from one column to the other. That is what the tables above are recording.

Two things that do hold

September is the weakest month everywhere on this list. It has the lowest typical figure at all five ski areas and was the snowiest month in only one winter of twelve at each. It is a real month to ski — the base is often at its deepest — but not a month to expect new snow.

Cerro Catedral is the most dependable July. Bariloche sits at 41° south, well below the central Andes, and its July was the biggest month six times out of twelve and never fell below 31 cm. Its snow is wetter — it sees rain on 30% of winter days — but for new snow in a fixed July week it is the safest bet here.

What to do with this

Do not book a fixed week in June, July or August expecting it to snow that week. On this record, a July at Valle Nevado or Portillo has about a one-in-three to one-in-four chance of delivering less than 20 cm.

Choose the high, cold ski areas, because they keep what falls. Snow at Valle Nevado falls at an average of −5.3 °C and sits at 3,000 m, so a big June can carry a dry July. At the lower volcanoes further south, a dry month is often a rained-on month as well. We measure snowfall rather than snow on the ground, so treat this as reasoning from the numbers rather than a measurement of it.

Later in the season usually means more snow on the ground, for the same reason February is Japan’s deepest month: it has had the most winter behind it. Late July and August trade a lower chance of fresh snow for a better chance of an established base.

Keep the booking flexible if you can. The three ski areas above Santiago are a day trip from the city, which is the one place in the southern hemisphere where you can wait for a storm and go the morning after.

How this was measured

Daily snowfall from the ERA5 reanalysis, summed by calendar month for each of the twelve most recent southern winters (June to September, 2014 to 2025), at each ski area’s own coordinates, on Japanese calendar days — the same day boundary as the rest of the site. Updated 13 September 2026: the first version counted UTC days, which moved several monthly figures by a few centimetres and Valle Nevado’s July 2024 from 2 cm to 1 cm; the conclusions are unchanged. The five ski areas were chosen as the best-known in each part of the Andes and sit in separate grid cells, so they are five measurements rather than one repeated. The Japanese and Alpine rows use December to March over the same twelve years.

“Typical” is the median winter, not the average, because the average is exactly what a few extreme months distort. The 20 cm line is our choice — roughly one proper storm — not a standard.

These are single winters, not averages, so nothing here depends on a short record smoothing out. That matters to us: an earlier version of our monthly figures, averaged over only three winters, showed a July collapse across the whole of South America that turned out to be an artefact of those three winters. Counting each winter separately is how we caught it.

The model understates mountain snowfall. Against 63 Japanese ground stations it returns about 56% of what they record, and we have no equivalent check for the Andes. Every centimetre here is likely a floor. The comparisons between months and between winters are the robust part, because they are made within the same place and the same model.