Japan Cherry Blossom 2027: Full Bloom Comes 7–10 Days After the Date You Looked Up

Published 28 July 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

Japan Cherry Blossom 2027: Full Bloom Comes 7–10 Days After the Date You Looked Up

Japan records two separate cherry blossom dates every year, for every city. Bloom — 開花 — and full bloom — 満開.

Most English-language guides quote one number and call it “peak”. It’s usually the first one. In Kyoto those two dates are, on average, nine days apart.

Nine days is not a rounding error. It’s most of a trip. It’s the difference between photographs of blossom and photographs of a tree with some optimistic buds on it.

I read the Japan Meteorological Agency’s own observation manual and its published averages for this, in Japanese, rather than other English guides — partly because the definitions are precise and stated only there, and partly because the organisation everyone credits with the cherry blossom forecast stopped producing one in 2010.

Jump to what you need:

The two dates, and why the gap costs people their trip

These aren’t informal terms. The JMA defines both precisely, in its Phenological Observation Guidelines, and observes them against a single designated tree in each city.

Bloom (開花):

標本木に５〜６輪の花が咲いた日を開花日とする

“The bloom date is the day on which five or six blossoms have opened on the sample tree.”

Five or six. On one tree. That is what the bloom date means — and it is emphatically not what a park looks like on that day.

Full bloom (満開):

咲き揃ったときの約80％以上が咲いた状態となった日を満開日として観測する

“The full bloom date is observed as the day on which about 80% or more of the blossoms that will open have opened.”

Note the wording carefully, because it’s easy to get wrong: it’s 80% of the blossoms that will eventually open, not 80% of the buds, and the guidelines add that they don’t all need to be open simultaneously — flowers that have already fallen still count.

One detail that tells you how precise this is: blossoms growing directly from the trunk or roots rather than the branches — 胴咲き — are explicitly excluded from the five-or-six count, because they open through a different process. That rule appears in the observation manual and essentially nowhere in English.

So which one do you want to be there for?

Full bloom, and the few days after it. Full bloom is the first day the tree crosses 80%; it stays worth seeing for several days afterwards.

The JMA does not define a “best viewing period” at all, so nobody can give you an official one. What its local offices do say is how long the transition takes. Shimonoseki’s district office, in its 2026 bloom announcement:

開花から満開となるまでの期間は、およそ１週間から10日を要します

“It takes approximately one week to ten days from bloom to full bloom.”

That matches the national picture. Which brings us to the numbers.

When each city blooms, and when it peaks

These are the JMA’s official averages, based on 1991–2020 data. Ordered by bloom date, which is roughly south to north.

(The gap column is my own subtraction of one JMA average from the other. The agency publishes both dates; it doesn’t publish the interval as a statistic.)

Two things worth pulling out of that table.

Kyoto and Tokyo do not peak together. Tokyo averages full bloom on 31 March, Kyoto on 4 April. Four days apart, in the same direction every year — which means a Tokyo-then-Kyoto itinerary in that order is working with the season, and the reverse order is working against it. If you have to choose, start in Tokyo.

The northern cities have less margin. In Kyoto you have nine days of transition to land in. In Sendai or Sapporo it’s five. If your trip is aimed at Tōhoku or Hokkaidō, your window is genuinely tighter and there’s less room for the season to move under you.

And it moves. A lot.

Averages are the right thing to plan on this far out, but you should know how wide the spread actually is. Tokyo’s last six years, observed:

2023 bloomed on 14 March. 2024 bloomed on 29 March. Fifteen days apart, in consecutive years, in the same city. Tokyo’s earliest recorded bloom is 14 March and its latest is 11 April — a month of range.

Plan your window on the averages. Do not book anything non-refundable on them.

Nobody at the JMA is forecasting this

Here is the thing almost every English-language cherry blossom article gets wrong, including some very large ones.

The Japan Meteorological Agency does not forecast cherry blossoms and has not done so since 2010. It announced the end of the programme on 25 December 2009:

気象庁でこれまで行ってきた応用気象情報としてのさくらの開花予想の発表につきましては、来春から行わないこととします。なお、生物に及ぼす気候の影響を知ることを目的としたさくらの開花の観測は引き続き行うこととします。

“The Agency will discontinue its cherry blossom forecast from next spring. Observation of cherry blossom flowering, for the purpose of understanding climatic effects on organisms, will continue.”

It had been forecasting since 1955. The last forecast was spring 2009. The reason given was that private weather companies had come to provide equivalent nationwide coverage.

So when a guide tells you “the JMA forecast says”, it is describing something that has not existed for over fifteen years. What the JMA does do — and does authoritatively — is observe and record. Every average and every observed date on this page comes from that record. It is the source everyone else is downstream of.

The forecasts come from private companies. Three matter, and their first releases land on a predictable schedule:

They then revise every week or two through February, and from March the JWA publishes full-bloom forecasts weekly alongside the bloom forecasts.

There is no 2027 forecast yet

I checked all three on 28 July 2026. None has published anything for 2027; their sites are still showing the closed-out 2026 season. The JWA’s page says so explicitly — the 2026 information is final and 2027 may differ.

So: any site showing you specific 2027 cherry blossom dates right now is showing you last year’s numbers. Based on the pattern above, expect the first real 2027 forecast in early January 2027, with the major ones landing through late January. That’s roughly five months from now.

I update this page when the forecasts are published. Nothing on it is a 2027 prediction, because there isn’t one to report.

What to book now, and what to wait for

The two halves of this trip have opposite deadlines, and getting that backwards is the expensive mistake.

Book now: accommodation.

Cherry blossom season is the tightest accommodation market in the Japanese calendar, and the good rooms in Kyoto and central Tokyo go six to twelve months out. You are currently inside that window for spring 2027 and near the front of it. Waiting for January’s forecast to book a hotel means booking after everyone else.

The way to square that with a forecast that doesn’t exist yet is straightforward: book the dates you want on free-cancellation rates, then adjust in January when the first forecasts land. You give up a small premium over the non-refundable rate. What you buy with it is the ability to move when the season moves fifteen days, which it demonstrably can.

(On what the rest of it costs: every fare and tax on a Tokyo–Kyoto week, sourced — including why nobody can honestly tell you the hotel figure.)

What to filter for: free cancellation, and a date range rather than fixed dates — Tokyo’s average full bloom is 31 March, so a stay covering roughly 27 March to 3 April brackets it. Central areas sell out first at this time of year, and rates rise steeply as the forecasts firm up. Which part of Tokyo, and why the answer starts with your airport — and note the hotel tax there changes on 1 April 2027, which is inside this window.

Kyoto peaks about four days after Tokyo — 4 April on average — so if you’re doing both, the Kyoto half wants to be the later half. Kyoto has far fewer rooms than Tokyo and they go earlier; it’s the harder of the two to fix late. Kyoto’s accommodation tax and which area actually works — it runs to ¥10,000 per person per night at the top end, which matters more here than in Tokyo.

Wait: fixed-date experiences, and anything you’d regret being early for.

Night illuminations, boat trips, guided walks and picnic spots are all worth doing and none of them needs booking in July. Several run on dates set only once the season is clear. Book these in January or February, once you know what you’re aiming at.

Come back to this in January, not now. Most blossom-season experiences aren’t loaded this far ahead, and the ones that are will be repriced once the forecasts land. The exceptions worth watching early are the things that are capacity-constrained regardless of blossom — tea ceremonies, cooking classes, anything with a small fixed group size. One thing genuinely not worth buying: entry to the famous parks and riversides. Ueno, Meguro, Maruyama, the Philosopher’s Path — they’re free, they’re public, and no ticket improves on turning up.

Tokyo’s hotel tax changes on 1 April 2027

This is the detail that makes spring 2027 different from any other cherry blossom season, and it is not yet in the English-language coverage at all.

Tokyo’s accommodation tax changes on 1 April 2027 — from a flat ¥100 or ¥200 per person per night to 3% of the room-only rate. The revised bylaw passed the Metropolitan Assembly and received central government consent on 30 June 2026.

Now put that next to the table above. Tokyo’s average full bloom date is 31 March.

The tax changes the day after Tokyo, on average, peaks.

What follows from that, concretely:

• The date that decides your rate is the date you stay, not the date you book. Booking in 2026 does not lock in the old rate. That much is settled in the ordinance’s transitional provision
• A typical blossom trip straddles the change. Arrive 27 March, leave 4 April, and part of your stay falls on each side
• Budget rooms get cheaper. The exemption threshold rises from ¥10,000 to ¥13,000, so a ¥12,000 room goes from ¥100 to nothing
• What I can’t tell you is how the Bureau treats a single booking spanning the changeover. They haven’t published a worked example. The tax is assessed per person per night and operators keep records by date of stay, so night-by-night is the natural reading — but that’s my reading, not their published position

The full breakdown, including how the tax is calculated when you share a room, is in the Tokyo guide. Kyoto and Osaka are unaffected — their rates don’t change in 2027.

It’s getting earlier

Worth knowing if you’re working from an older guidebook, or from what a friend did a decade ago.

The JMA’s own climate change assessment states the trend plainly:

さくらの開花日は、10年当たり1.2日程度早くなっている

“Cherry blossom bloom dates are becoming earlier by approximately 1.2 days per decade.”

That’s measured across 58 stations from 1953 to 2024, and the agency attributes it to long-term temperature rise. You can see it in the averages themselves — when the JMA moved from the 1961–1990 baseline to the current 1991–2020 one, Tokyo’s average bloom shifted from 29 March to 24 March. Five days earlier, between two official averages.

Most major cities moved four to six days earlier over the same comparison. Which means: advice written before about 2010 runs late, and the older the source the later it runs.

Except in Okinawa, where it’s going the other way

Naha’s average bloom date did not move at all between the two baselines — 16 January in both. Ishigaki’s went three days later.

Okinawa’s blossom isn’t the same tree. Yoshino cherry can’t grow there, so the JMA observes 寒緋桜 instead, and it has different cold requirements — it needs a proper chill to trigger flowering, so warmer winters delay it rather than advancing it.

It’s a small thing, but it’s a good check on any source you’re reading: if a guide tells you Japan’s blossom is uniformly moving earlier, it hasn’t looked at the southern end.

Quick answers

Is Sapporo’s blossom a different species? No — Sapporo is observed on Yoshino cherry, the same as Tokyo and Kyoto. This is a common mistake in English guides. Other Hokkaidō stations are different: Asahikawa, Abashiri, Kushiro, Obihiro and Wakkanai are observed on Ezo mountain cherry, and Nemuro on Chishima cherry. But Sapporo, Muroran and Hakodate are Yoshino.

Where is the tree they measure? It varies by city, and the JMA only names it in some places. Tokyo’s has been at Yasukuni Shrine since 1966; Osaka’s has been in Osaka Castle Park’s Nishinomaru Garden since 1977. Both are confirmed by the relevant regional observatory. For most other cities I could not find the location stated in an official source, so I’m not going to repeat what unofficial sites say — and note that sample trees do get replaced, so even a correct answer can go stale.

How far ahead can a forecast actually be trusted? The early-January releases are directional. They firm up substantially through February as real temperature data accumulates. If you need to commit to non-refundable arrangements, the February revisions are a much safer basis than the January ones.

Is cherry blossom season worse for crowds than autumn? It’s the tightest accommodation market of the year, which is a different thing from the worst crowds. Kyoto’s late-November foliage weeks concentrate visitors harder at individual temples. Spring spreads across far more of the country because the blossom itself does. The autumn comparison is here.

Should I buy a rail pass to chase the blossom front north? Almost certainly not, unless you’re genuinely covering long distances in a short time. The blossom front moves slowly enough that most trips see one or two regions, not five. Run the numbers before you buy anything.

What if I can only travel in mid-April? Go north. Sendai and Niigata average full bloom on 13 April, Kanazawa on 8 April. Mid-April is a bad date for Kyoto and a good one for Tōhoku — the season hasn’t ended, it’s moved.

What to do next

1. Pick your window from the full bloom column, not the bloom column — and give yourself a few days either side
2. Book accommodation now, on free-cancellation rates. This is the part with a real deadline
3. Put a reminder in early January 2027 for the first Weathernews forecast, and late January for the JWA’s
4. Adjust your dates then, while cancellation is still free
5. If your stay crosses 1 April 2027, add 3% of the room-only rate to your Tokyo nights when comparing prices → the Tokyo guide
6. Book fixed-date experiences in January or February, once the forecast is real

Bloom and full bloom averages, observed dates, definitions and the species assignments on this page come from the Japan Meteorological Agency’s published records and its Phenological Observation Guidelines, read in Japanese. Averages are the 1991–2020 baseline. Forecast release dates come from the forecasters’ own announcements. The Tokyo accommodation tax figures come from the Tokyo Metropolitan Bureau of Taxation and the revised bylaw. Where I couldn’t verify something at source — the location of most sample trees, and how a stay spanning 1 April 2027 is taxed — I’ve said so rather than filling the gap.

Last checked 28 July 2026.