· updated
Japan's Snowiest Hot Spring Town Has Two Places to Stay. Its Most Famous One Ranks 772nd.
Otari Onsen sits in 690 cm of snow a season, the deepest of Japan's 79 designated onsen towns. Ranked by measured snowfall, with how many inns each town actually has and how far it is from the nearest city. No skiing required.
You do not have to ski to want this. Snow falling on a wooden bathhouse, steam coming off the water, and the specific quiet that arrives when a metre of snow has landed on a village overnight.
The question nobody answers with numbers is which hot spring towns actually get that snow. Plenty of lists tell you which onsen are famous. Fame and snowfall turn out to be almost unrelated.
The ranking
We measured snowfall at 3,065 ski areas worldwide using one method, which gives us a like-for-like figure for every Japanese hot spring town that has a ski area attached, and nearly all the snowy ones do. Alongside it: how many places to stay sit within 6 km, and the nearest city of any size.
| Hot spring town | Season snowfall | World rank | Places to stay | Nearest city |
|---|---|---|---|---|
| Seki Onsen | 689 cm | 19 | 2 | Nagano, 29 km |
| Nozawa Onsen | 450 cm | =107 | 36 | Nagano, 39 km |
| Hakusan Ichirino | 510 cm | 66 | 2 | Kanazawa, 34 km |
| Togari Onsen | 450 cm | =107 | 1 | Nagano, 37 km |
| Akakura Onsen | 445 cm | =116 | 31 | Nagano, 27 km |
| Oze Hinoemata | 459 cm | 108 | 6 | Utsunomiya, 68 km |
| Amihari Onsen | 358 cm | =261 | 1 | Morioka, 21 km |
| Hirayu Onsen | 384 cm | =199 | 46 | Matsumoto, 37 km |
| Onogawa Onsen | 314 cm | =426 | 53 | Fukushima, 39 km |
| Asarigawa Onsen | 293 cm | =502 | 6 | Sapporo, 26 km |
| Zao Onsen | 258 cm | =670 | 56 | — |
| Kusatsu Onsen | 252 cm | =709 | 66 | — |
Two numbers in that table do most of the work.
Seki Onsen gets 689 cm a season, 19th in the world, and has two places to stay. Not two hotels and some guesthouses. Two.
Kusatsu Onsen, the one most people outside Japan can name, ranks 776th and has 66. Kusatsu is a genuinely great hot spring town with a famously strong sulphur spring. It is not a place you go for snow depth.
Why the two run in opposite directions
This is not a coincidence and it is not unfair. It is the same fact stated twice.
A town where five metres of snow lands every winter is a town that is hard to reach, hard to build in, and expensive to keep open. Roofs need clearing. Roads need ploughing all night. Nobody builds sixty inns in a place like that, because for four months a year the cost of simply existing there is enormous.
The famous onsen towns are famous partly because they were easy to get to — near a road, near a city, on a route people already travelled. Easy to get to and buried under snow are close to mutually exclusive in Japan.
So the honest framing is a trade, not a ranking:
- Maximum snow means a village with one or two inns, no shops open in the evening, and no plan B if you do not like where you are staying.
- Maximum comfort means less snow, and often a great deal less.
Where the trade is best
Nozawa Onsen is the clearest answer for most people. 450 cm puts it 107th in the world, one centimetre off Whistler’s 511, and it has 36 places to stay and thirteen free public bathhouses run by the neighbourhoods that own them. It is a working village, not a resort, and it is the only place in this table that is both genuinely snowy and genuinely equipped.
Akakura Onsen is the runner-up on the same logic: 466 cm, 96th, 31 places to stay, and the closest of any of them to a major city at 27 km from Nagano.
Hirayu Onsen deserves a mention for a different reason. 362 cm is not top-tier by this table, but it has 46 inns and sits in the Okuhida valley, which is one of the few places in Japan where the outdoor baths look out at genuine alpine peaks rather than forested hills.
The part that surprises people
Look at the last column again. Almost every town in the top ten is under 40 km from a city of half a million people.
Seki Onsen is 29 km from Nagano. Akakura is 27 km. Nozawa is 39 km. Asarigawa is 26 km from Sapporo, a city of 2.5 million.
Japan’s deep snow is not remote in the way deep snow usually is. The Sea of Japan effect dumps it on the near side of the mountains, within an hour or two of cities that a bullet train already reaches. This is the single biggest practical difference between a Japanese winter trip and the equivalent in the Alps or the Rockies, and it is why a trip here works without a car and without a long transfer.
Nagano is about 90 minutes from Tokyo on the Hokuriku Shinkansen. From there, buses run to most of the names above.
How to choose
If you want the deepest snow and will accept anything else: Seki Onsen. Book early, because there are two options and they are not large.
If you want deep snow and want the trip to be easy: Nozawa Onsen or Akakura Onsen. Both are reachable from Tokyo in about three hours without a car, both have real inventory, and both are in the world’s snowiest hundred.
If you want the famous names: Kusatsu and Zao are excellent hot springs. Go for the water and the town. Do not go expecting the snow to be the deepest in Japan, because the data says clearly that it is not.
If you are going in a group with one skier and one non-skier: every town in this table has a ski area attached, which is why we have snowfall figures for it at all. That is the arrangement most of them are built for.
What we cannot tell you
We do not have data on which baths are outdoor. Every town here has hot spring water; whether a given inn has a rotenburo facing the snow is a property of the inn, not of the town, and we do not hold that. Check the inn.
The count of places to stay is a floor, not a total. It comes from OpenStreetMap within 6 km, which undercounts small family-run inns badly — exactly the kind of inn these towns are made of. Nozawa’s real number is higher than 36. The comparison between towns still holds, because every town was counted the same way.
A town is not its ski area. We locate each town by the ski area that carries its name, so the snowfall figure is measured at the slope rather than in the bath. In a valley this narrow the difference is small, but it is not zero.
How this was measured
Daily snowfall summed over each of the twelve winters from 2014/15 to 2025/26 and averaged per season, winter defined per hemisphere. Source is Open-Meteo historical data from the ERA5 reanalysis.
These figures are estimates, and they run low. ERA5 works on a 10–30 km grid, which flattens the mountain, so orographic uplift is under-produced and mountain snowfall comes out well below what actually falls.
We checked it against Japan Meteorological Agency stations at 156 Japanese ski areas, over the three winters those stations cover (2023/24 to 2025/26). The model figures in this paragraph are measured over that same three-winter window, so they are not the twelve-winter averages shown elsewhere on this site. It runs low everywhere, but not evenly — close on the Sea-of-Japan side of Honshu, about half in Hokkaido, where the colder, drier snow piles up deeper than a fixed precipitation-to-snow ratio can see. We do not correct for it, because only Japan can be checked this way and a Japan-only correction would distort every international comparison on the site.
What follows: narrow gaps here cannot be trusted, and wide ones can. Two towns a few dozen places apart are not distinguishable. A town at two or three times another is.
Nozawa, Hakusan Ichirino and Togari return the same 450 cm because neighbouring places share a grid cell. Treat them as indistinguishable rather than tied.
Distances to the nearest city are straight-line to a city of at least 100,000 people, not road distance. A dash means no city of that size falls inside the search radius — Zao Onsen sits inside Yamagata City itself and Kusatsu is deep in the Gunma hills, so neither produces a “nearest city” figure the way the others do.
Book the trip
We do not take bookings. These are the sites Japanese ski accommodation actually sits on.
Contains advertising. We may earn a commission on bookings made through Rakuten.
Keep reading
Other guides built from the same dataset.