Japan Autumn Foliage 2026: Nikkō Late October, Kyoto Into December

Published 27 July 2026 · Updated 5 October 2026

This article contains affiliate links. If you book through them I may earn a small commission at no extra cost to you. I only link to services I’d use myself. #ad

Most English-language guides to Japan’s autumn colours will tell you to visit in November.

That advice will make you miss it.

You’ve booked the flight. You picked November because every article says November. In more than half of Japan, November is the wrong month entirely.

Japan’s foliage season runs for roughly two and a half months — from the first week of October in the mountains of Hokkaido to the first week of December in southern Kyoto. Show up in Nikko in late November and you’ll find bare branches. Show up in Kyoto in late October and you’ll find green.

I live in Japan and follow the Japanese-language forecasts each year. This guide gives you the timing that actually matters — not the national average, the window for the specific place you booked.

Jump to what you need:

The one thing you need to understand first

Japan’s autumn front — kōyō zensen (紅葉前線) — moves in two directions at once:

1. North to south — Hokkaido first, Kyushu last
2. High elevation to low elevation — mountains first, cities last

This second one is what English guides almost always leave out, and it’s the more useful of the two.

In October, elevation matters more than latitude. A mountain in central Japan will turn before a city far to its north.

In late November, latitude matters more. By then the mountains are bare and the season belongs to the low-lying cities.

Practically, this means: if you’re visiting in October, go up. If you’re visiting in late November, stay low.

City-by-city peak timing

These are typical peak windows based on historical averages. Actual dates shift by roughly one to two weeks in either direction each year depending on summer heat and the timing of the first cold snap.

The two dates most trips get wrong:

• Nikkō peaks in late October, not November. Its most famous stretch — the Irohazaka switchback road — is one of the earliest major spots in central Japan because of elevation. Follow a guide that leaves that out, show up in November, and your one shot at Nikkō is bare branches, not colour.

🔗 See Nikkō day trips and Tōbu passes → Nikkō is the one autumn destination where the date matters more than the booking. It’s a day trip from Tokyo — direct limited expresses from Asakusa take about 1 hour 50 minutes, ¥3,340 each way — so you don’t need to build a trip around it. You need to be there in the right ten days. Get that part wrong, and nothing else on the itinerary makes up for it. The specific problem is Irohazaka. It’s a one-way pair of switchback roads up to Lake Chūzenji, and in peak foliage weekends it backs up badly enough to eat several hours — hours you won’t get back on a single day trip. If your dates land in late October, either go on a weekday, or take a coach with a reserved departure so the queue is somebody else’s problem. Outside those weeks the buses are fine and you should just use the Tōbu pass. If Nikkō is on your list, where you stay in Tokyo decides how easy this day is — the passes only work from one side of the city. - Kyoto runs later than people expect, and it runs long. The city’s official observation point has a climatological normal of 5 December and is forecast at 11 December for 2026 — but Arashiyama is forecast to begin peaking on 13 November, because they are different trees about three weeks apart (both forecasts are further down this page). So the window is wide on purpose: mid November through mid December, with the city’s own temples at the back of it. If your dates are early November, Kyoto is still not your destination — Hakone or Kawaguchiko is.

The first 2026 forecast is out — and Kyoto is later than my own table says

Japan’s meteorological companies publish their foliage forecasts on a fixed schedule, and there is more than one of them. The Japan Weather Association (日本気象協会) and Weathernews release a first forecast in September or early October, then revise it two or three more times through October and early November as the actual temperature data comes in. A third company, Japan Meteorological Corporation (日本気象株式会社), publishes on the same schedule — and this year it went first, on 2 September, with the Japan Weather Association following on 1 October. Note the first and third of those: 日本気象協会 and 日本気象株式会社 differ by one word and are separate companies. If you see a headline citing “Japan Weather” without a link, it could be either.

The first 2026 forecast is out. Japan Meteorological Corporation published its 紅葉・黄葉見頃予想（第1回）on 2 September 2026, covering around 700 viewing spots and 3,000 mountains. These are peak viewing dates — 見頃 — which is the same measure my table above estimates, so they are directly comparable. Where it disagreed with my table I have said so below and moved the row, rather than quietly leaving both numbers up.

Read the third column, not the second. JMC labels every one of those city rows 平年並 — within normal — except Sapporo and Nagano, which it calls late outright. So the headline is not “a late year.” It is a normal year in the cities most people are booking, and a late one in the north.

⚠️ That was one company’s first pass. Forecasts get revised two or three times before November, and a first round is the least reliable. Do not rebook around a single number six weeks out.

The Japan Weather Association has now published too — and it emphasises the opposite

The Japan Weather Association released its 2026年紅葉見頃予想＜第1回＞ on 1 October 2026. It forecasts named spots rather than cities, which is more useful for planning and harder to compare against JMC. These are the ones a visitor is likely to be choosing between:

🔴 There is a trap in that last column, and it is the reason to read the forecast rather than the table. The column compares 2026 with last year, not with the long-run average. Almost every row reads earlier. The Association’s own written outlook for the season says the opposite about the average: 全国的に平年並みか平年より遅くなる — nationwide, normal to later than average — and it singles out the lowlands of eastern and western Japan as the places running late. Both statements are true at the same time, because 2025 itself came in late. A guide that scrapes the table and skips the prose will tell you 2026 is an early year. The Association is not saying that.

This restores advice I had removed. On JMC’s September round I wrote that the “weight your plans toward the later end” guidance applied to the north and not to the Tokyo–Kyoto corridor. The Association’s first round puts eastern Japan — 関東甲信 and 東海 — at 平年並みか平年より遅く, normal to later. And the single Tokyo-area spot it publishes, Mount Takao, is one of only two rows in its entire national table called later than last year — and by far the larger gap of the two, eight days against one day for Yahiko Park in Niigata. Every other spot in the country is called earlier than last year or the same. So the later-end advice is back on for the corridor.

Two companies, two first rounds, opposite emphasis. That is the honest state of the 2026 forecast in early October. I cannot resolve it for you, and neither can anyone quoting just one of them.

🔴 The September forecast moved my table, and the October one moved part of it back. Kyoto used to read mid November – early December on this page. Two things said that was too early for this year:

1. JMC forecasts Kyoto’s maple peak at 11 December — past the end of “early December” on any reading.
2. Kōdai-ji is running its 2026 evening viewing through 13 December (dates further down this page). A temple staffing and ticketing an evening opening that late is making a commitment, not a guess.

And one thing that does not support the change, which you should weigh against the two above. The long-run averages are fine with my old window: JMC’s climatological normal for Kyoto is 5 December, and the Japan Weather Association’s normal for Arashiyama is mid-November to early December. Both sit inside what this page already said. So this is a 2026-specific adjustment, not a correction of a long-standing error — in a typical year, “mid November to early December” would still be the right thing to tell you.

So in September I moved Kyoto to late November – mid December. Tokyo (forecast 29 Nov, normal 28 Nov) and Osaka (4 Dec) both sat inside their existing windows and were unchanged.

🔴 The Japan Weather Association’s October round then put the front edge back, and I have widened the row to mid November – mid December. Its Arashiyama figure is 13 November, called the same as last year — about ten days before my September window even opened. The two companies are roughly four weeks apart on Kyoto, and the reason is that they are not measuring the same tree.

Neither is wrong, and the gap is not a 2026 effect — those two normals are about three weeks apart in an ordinary year. Kyoto’s answer depends on which Kyoto you mean. Arashiyama turns weeks before the tree the city’s official date is read from.

What that costs this page’s central claim, stated plainly: sharpening Kyoto to a December window on one company’s first round was too confident, and a second company’s first round was enough to widen it again. Colour at Arashiyama from the middle of November is a real forecast from a real source. Kyoto still runs later than the English-language guides say, which is the claim that matters. “Kyoto is a December destination” was further than the evidence went.

If you already booked Kyoto on the old window, the last week of November is not wrong — it is the front of the range rather than the back of it. But if foliage is the reason for the trip, I would not now book Kyoto dates that end before 5 December.

⚠️ The species is in the forecast and it matters. JMC’s Kyoto, Osaka and Tokyo figures are for いろはかえで (Japanese maple). Sapporo’s is やまもみじ, a different tree on a different clock. A city’s “peak” is a peak for the species that city is measured on.

The rest of my table is built on historical averages; the forecast above is one company’s first projection for this specific year. Averages are the more stable of the two this far out, which is why the other rows still show them — I moved Kyoto on this year’s evidence rather than on the averages, which is the weaker kind of reason — so treat that row as the one most likely to move again. I update this page each time a new forecast is released — bookmark it if you’re travelling this autumn.

Spring works the same way, with one difference worth knowing. The cherry blossom forecasts come out in January, not September — and they don’t come from the Meteorological Agency at all, which stopped forecasting blossom in 2010 and now only records it. The spring timing guide is here, including the detail that trips up more people than anything in autumn: bloom and full bloom are two different official dates, seven to ten days apart.

One pattern worth knowing: recent years have consistently run later than average. The mechanism, in the Japan Weather Association’s own words, is that peak comes earlier when the autumn runs cold and later when it runs warm — September through November, not the summer. That’s also why the forecast gets revised two or three times: each round folds in the temperature that has actually arrived. The 2025 season came in “average to slightly late” nationwide. I wrote here that if the trend held, you should weight your plans toward the later end of each window above.

The two first rounds split on this, and the later one is the one to follow. JMC’s September round labelled Tokyo, Osaka and Kyoto within normal and called only Sapporo and Nagano late, which is why this paragraph briefly said the later-end advice applied to the north only. The Japan Weather Association’s 1 October round says 全国的に平年並みか平年より遅くなる — normal to later nationwide — and names the lowlands of eastern and western Japan as the late ones, for a stated reason: September’s second half ran warm and October and November are forecast to keep running warm. It is a month newer and it has more of the autumn’s actual temperature in it. Weight toward the later end, corridor included.

Book accommodation before you finalise anything else

This is the part that catches people out.

During peak foliage weeks, hotels in Kyoto, Nikkō, Hakone and Takayama don’t just get expensive — they sell out. Kyoto in the last two weeks of November is one of the tightest accommodation markets in the entire Japanese calendar, comparable to cherry blossom season. By the time the forecast confirms your dates, there may be nothing left in the city at all.

If Kyoto is already gone for your dates, it isn’t over. Osaka is about half an hour away by fast train, has far more rooms, and caps its accommodation tax at ¥500 where Kyoto’s runs to ¥10,000. Basing in Osaka and day-tripping into Kyoto is a legitimate plan in peak weeks rather than a consolation prize — you lose the early mornings, which is a real cost, and you gain availability at a sane price.

The sequence that works:

1. Pick your week using the table above
2. Book accommodation immediately — it’s the constraint
3. Book transport and activities after

Not the other way around.

Where to book, city by city. Which area you choose matters more in these weeks than at any other time of year, because the well-connected rooms go first and the ones left are the ones you’d need a bus to reach. Kyoto — which area, and the tax that changes the bill · Osaka — the base that stays available · Tokyo — start with your airport For Hakone and Kawaguchiko I don’t have a guide worth linking to yet. Book those direct and early; the lake-view rooms go first.

A note on free cancellation: normally I’d tell you to take the cheaper non-refundable rate. Not this time. The first 2026 forecast is now out, but it is one company’s opening projection and it gets revised two or three more times before November. Paying a few dollars more per night to keep your dates flexible into November is the right trade here — and it’s cheaper than rebooking, which is the arithmetic of where Japan trips actually overspend.

Tokyo nights are the exception to the urgency. Tokyo has far more rooms than Kyoto or Nikkō and doesn’t sell out the same way, so book the constrained places first and Tokyo last. Where in Tokyo, and why the answer starts with your airport — and if Nikkō’s late-October peak is what you’re chasing, the Asakusa end of the city is where the direct trains leave from.

Before you book Kyoto, one thing that changes the bill

Kyoto raised its accommodation tax on 1 March 2026. It’s charged per person, per night, and the top bracket went from ¥1,000 to ¥10,000 — a tenfold rise most English guides still haven’t picked up.

The part that saves you money: it’s assessed on the room-only rate, with meals excluded. Ryokan almost always quote a rate with dinner and breakfast included — which is what you’ll be looking at in Hakone, Kawaguchiko and Takayama — so the taxable figure is lower than the price you’re quoted, sometimes by a whole bracket.

You’re booking peak season at peak prices. This is worth ten minutes first:

→ Where to stay in Kyoto in 2026, and what the tax will actually cost you

Evening illuminations: the part that needs planning

Many temples and gardens run night-time illuminations (ライトアップ) during foliage season. They’re spectacular, and they are handled very differently from daytime visits:

• They run for a limited window — often just two to three weeks, tied to the local peak
• Several major Kyoto temples now require advance reservation or timed-entry tickets
• Dates are frequently announced only on Japanese-language pages, sometimes as late as October

This is genuinely hard to research from outside Japan, and it’s where most English itineraries fall down. Reservation systems for the popular Kyoto illuminations can fill within days of opening.

If you’re planning around illuminations, check the individual temple’s Japanese site rather than relying on aggregator listings, which are often a year out of date. I’ll add confirmed 2026 illumination dates to this page as they’re announced.

Confirmed 2026 dates, as they land

Kōdai-ji (高台寺), Higashiyama — 23 October to 13 December 2026. The temple’s own Japanese homepage carries the 夜間特別拝観 window in a single line, and it was already up when I checked on 8 September, earlier than the “late September at best” that this section used to assume.

That end date is worth more than it looks. Kōdai-ji is not guessing about foliage in the abstract — it is committing staff, lighting and a ticketing system to being worth visiting after dark through 13 December. Set that beside JMC’s forecast of 11 December for Kyoto’s maple peak, and you have two independent parties, one a forecaster and one a temple with money at stake, landing in the same week. Neither is affiliated with the other. That is the strongest evidence on this page that Kyoto’s autumn is a December event, and it is why the Kyoto row above now runs to mid-December instead of early December.

I will add temples here as they publish. Most have not yet.

🔗 Browse Kyoto illumination tickets → Where advance tickets are sold to overseas visitors, this is usually the easiest route — the direct Japanese reservation systems often require a Japanese phone number.

Beating the crowds

Autumn is Japan’s second-busiest tourist season after cherry blossom, and the famous spots are genuinely crowded — the photo you’re picturing was taken by someone who got there before you did.

Three things that actually work:

Go at opening time. Most major Kyoto temples open at 8:00 or 9:00. The difference between arriving at opening and arriving at 11:00 is not subtle — it’s the difference between a photograph and a queue. Arashiyama in particular is transformed.

Go on a weekday. Japanese domestic tourism concentrates hard on weekends during foliage season. A Tuesday at a famous temple can be half the density of the Sunday before it.

Go one stop further. The single most crowded spots are famous partly because they’re easy to reach. Northern Kyoto (Takao, Kurama, Kibune) peaks earlier than central Kyoto and takes a fraction of the visitors. Same season, different experience.

Takao is the sharper case of the two. Weathernews forecasts Jingo-ji roughly a week ahead of the Higashiyama temples, and its own note on that temple is that it colours earlier than the rest of the city; Kurama and Kibune run two or three days ahead rather than a week. That is one temple’s forecast rather than a reading of the whole valley — Saimyō-ji and Kōzan-ji are not on Weathernews at all — so take it as a reason to go early in the Takao window, not as a date for the three temples together.

Sample timing for a 10-day trip

If your dates are late October: Nikkō (peak) → Takayama (peak) → Kanazawa (early colour). Skip Kyoto — you’ll be a month early.

If your dates are early-to-mid November: Hakone or Kawaguchiko (peak, with Fuji views) → Karuizawa (peak) → arrive Kyoto at the very end as it starts turning.

If your dates are late November: Kyoto (peak) → Nara (peak) → Tokyo (turning). This is the classic route, and it’s the classic route for a reason — you’ll also be sharing every temple with everyone else who read the same advice.

If your dates are early December: Kyoto’s final week → Tokyo ginkgo avenues → Kamakura. Everyone else has already gone home, convinced it’s over. It isn’t. This is often the best combination of colour and space in the entire season — you’re seeing it without the crowd that already left.

Before you buy a rail pass for any of these: three of the four routes above don’t come close to justifying a nationwide JR Pass, and the fourth is borderline. A Tokyo–Kyoto round trip covers barely half its cost. Run the numbers here first — for most autumn itineraries a regional pass is cheaper, often by a lot, and the difference is roughly the price of two extra nights.

FAQ

Does the foliage season shift much year to year? Yes — typically by one to two weeks. A warm autumn pushes it later; the forecasters key it to September–November temperatures, not to the summer. Plan a window, not a date, and check the September forecast before committing to anything non-refundable.

Is autumn more crowded than cherry blossom season? Slightly less at the national level, but the concentration is comparable at the famous Kyoto temples in late November.

What about Tokyo? Tokyo’s autumn is mostly ginkgo rather than maple — yellow rather than red — and it peaks late, from the last days of November into early December. The ginkgo avenues are a genuinely different look from the temple maples, and worth seeing on their own terms rather than as a consolation prize.

Since almost everyone starts or finishes here: which part of Tokyo to book depends on your airport more than your neighbourhood, and if Nikkō is on your list, one specific area makes that day trip much easier.

Can I see foliage and snow monkeys / early skiing in one trip? Late November into December, yes. Higher-elevation Nagano starts getting snow while the lowland cities are still in colour.

What to do next

1. Pick your week from the table above
2. Book refundable accommodation now — it’s the binding constraint, and it only gets tighter
3. Sort out where in Kyoto, since that’s where the money and the crowding both concentrate → the area guide, and the tax that changed in March
4. Then, and only then, work out transport → whether a JR Pass is worth it for your route (for most autumn trips: no)
5. Two first-round forecasts are now in (JMC on 2 September, the Japan Weather Association on 1 October) and both sets of dates are above. Check back through November — each gets revised two or three more times, and the two currently disagree about Kyoto by four weeks

The timing table here is built from Japanese-language forecast data and historical averages, not from other English guides — which is why it doesn’t say “visit in November.” I said on this page that I would update it when the Japan Weather Association published its 2026 round. It published on 1 October and this page changed the same week, including widening a row I had narrowed three weeks earlier.

Forecast status, checked 5 October 2026: Japan Meteorological Corporation published its first 2026 round on 2 September; the Japan Weather Association published its first on 1 October. Both sets of figures are above, including where they disagree. Weathernews has a live spot page for 2026 but no dated first round of its own that I can cite, so nothing from it is quoted here. Any site still showing you a single confident 2026 Kyoto date is not reading both of the forecasts that exist.