Twelve castle keeps in Japan survived the Meiji demolitions, the fires and the war. Every other keep you see standing today — Osaka, Nagoya, Okayama — is a reconstruction. That much is settled. What is not settled is how many people visit those twelve, because the answer is scattered across twelve different Japanese-language statistical reports, each of which counts something slightly different. This page pulls all twelve together from primary sources, with the year and the counting unit attached to every figure, and then explains why you should not turn the result into a ranking.
Cite this page: Tabilane. "Japan Castle Statistics: Visitor Numbers for All 12 Original Keeps (2026)." tabilane.com. https://tabilane.com/japan-castle-statistics
Quick Answer
Twelve original castle keeps (genson tenshu) survive in Japan. Five are National Treasures — Matsumoto, Inuyama, Hikone, Himeji and Matsue — and the other seven are Important Cultural Properties.
Visitor figures exist for all twelve, but they are not comparable with one another. Only five of the twelve come from actual paid-admission counts; the rest are visitor-days, municipal estimates or aggregated report totals, and the years range from 2023 to fiscal 2025. The largest verifiable ticketed figure is Himeji Castle at 1,567,674 paid admissions in FY2025.
Ask "which Japanese castle gets the most visitors?" and the honest answer is that it depends which turnstile you count. Matsuyama City publishes three different numbers for Matsuyama Castle in the same document for the same year — 53,400, 612,203 and 1,638,438 — and none of them is wrong. Shimane Prefecture lists Matsue Castle twice in the same table under the same classification code, then adds both rows together in its own summary.
Himeji Castle. Photo: Gorgo via Wikimedia Commons (public domain).
The 12 original keeps — and what the designation data actually says
The twelve are Hirosaki, Matsumoto, Maruoka, Inuyama, Hikone, Himeji, Matsue, Bitchū-Matsuyama, Marugame, Iyo-Matsuyama, Uwajima and Kōchi.
The five National Treasure keeps were designated as follows, according to the Agency for Cultural Affairs' national heritage database:
| Castle | Designated | Period / date given |
|---|---|---|
| Himeji (main keep) | 9 June 1951 | Momoyama / 1608 |
| Matsumoto | 29 March 1952 | Early Edo / c. 1615 |
| Inuyama | 29 March 1952 | Momoyama / 1601 |
| Hikone | 29 March 1952 | Momoyama / 1606 |
| Matsue | 8 July 2015 | Momoyama / 1611 |
Three of the five were designated on the same day in 1952, Matsumoto Castle among them. The fifth, Matsue, joined sixty-three years later, in 2015 — a detail that English-language coverage of "Japan's five National Treasure castles" almost never mentions.
The other seven are Important Cultural Properties: Maruoka, Uwajima and Kōchi (all 30 January 1934), Iyo-Matsuyama (1935), Hirosaki (1937), Bitchū-Matsuyama (1941) and Marugame (1943). So eleven of the twelve were designated between 1934 and 1952, and one arrived in 2015.
There is a subtlety worth knowing before you quote the number twelve. The designation records do not use a single consistent unit. Some entries list a keep on its own ("Inuyama Castle Keep"). Some nest it inside a parent designation ("Matsumoto Castle Keep — Keep", which also covers the adjoining turrets). Hikone's designation reads "Hikone Castle Keep, Attached Turret and Multi-Bay Turret." Himeji has more than one National Treasure keep entry, because the great keep and the smaller Inui keep are listed separately. Kōchi's buildings are itemized one by one.
None of this makes "twelve original keeps" wrong — it is the standard, widely agreed framework in Japan. It does mean that the number twelve is the product of a particular way of counting, not something you can read straight off the national register. That distinction turns out to be the theme of this entire page.
Visitor numbers for all 12 castles
Every figure below comes from a prefectural or municipal statistical publication, retrieved in August 2026. The year column and the "what is counted" column are not decoration — without them the table is meaningless.
| Castle | Prefecture | Figure | Year | What is counted |
|---|---|---|---|---|
| Himeji | Hyōgo | 1,567,674 | FY2025 | Paid admissions (ticketed) |
| Matsumoto | Nagano | 986,200 | 2024 | Visitor-days |
| Hikone | Shiga | 636,289 | 2024 | Tourist arrivals, aggregated municipal reports |
| Iyo-Matsuyama (keep) | Ehime | 612,203 | 2025 | Keep admissions (ticketed) |
| Inuyama | Aichi | 582,447 | 2023 | Tourism and recreation users |
| Matsue | Shimane | 402,049 | 2024 | Visitor-points |
| Hirosaki (honmaru area) | Aomori | 356,811 | 2024 | Paid-area entries |
| Kōchi (Kaitokukan) | Kōchi | 272,037 | 2024 | Facility users |
| Maruoka | Fukui | 203,000 | 2024 | Visitor-days, rounded to thousands |
| Marugame (keep) | Kagawa | 122,583 | 2024 | Aggregated report, per person |
| Uwajima | Ehime | 57,915 | 2024 | Reported estimate |
| Bitchū-Matsuyama | Okayama | 56,266 | 2024 | Paid-facility admissions |
Please do not read this as a ranking. It is sorted by magnitude because a table has to be sorted somehow, but the twelve rows span three calendar years plus one fiscal year, and at least four distinct counting units. Comparing 1,567,674 ticketed admissions in Himeji with 986,200 visitor-days in Matsumoto is not a comparison — visitor-days count the same person more than once by design.
Only five of the twelve are counts of paid admissions: Himeji, Iyo-Matsuyama, Marugame, Bitchū-Matsuyama and Hirosaki. The remaining seven are visitor-days, estimates or aggregated municipal returns. They are not measuring the same quantity.
One naming note, because it causes real confusion in English: there are two Matsuyama Castles. Iyo-Matsuyama is in Ehime on Shikoku; Bitchū-Matsuyama is a mountaintop castle in Okayama, on the far side of the Inland Sea. Japan's own designation records list the Ehime one simply as "Matsuyama Castle," with no prefix. Always take the prefix.
Why you can't rank that table: four ways the count splits
The gap between two numbers for the same castle is usually larger than the gap between two different castles. There are at least four distinct mechanisms, and they show up inside single documents.
1. How far up you went
Matsuyama City's 2025 tourism estimates publish three separate figures for the castle on Mount Katsuyama, in the same PDF, for the same year:
| What you paid for | 2025 |
|---|---|
| Ninomaru Historical Garden admissions | 53,400 |
| Castle keep admissions | 612,203 |
| Ropeway and chairlift passengers | 1,638,438 |
That is a 30.7× spread between the smallest and the largest, and all three are correct. Some people stroll the garden at the foot of the hill. Some ride up and walk the grounds without paying to enter the keep. Some buy the keep ticket. The ropeway figure also excludes anyone who walked up, so it is neither a ceiling nor a floor for visits to the castle hill — which stands a short tram ride from Dōgo Onsen, the reason most travelers are in Matsuyama at all.
Iyo-Matsuyama Castle keep. Photo: Mnd via Wikimedia Commons, CC BY-SA 3.0.
2. Everyday admission versus the festival
Aomori Prefecture's 2024 statistics list the paid area of Hirosaki Castle at 356,811 — and, on a different page, the Hirosaki Cherry Blossom Festival at 2,450,000, the largest single event in the prefecture. Same park, 6.9× apart, counted under different headings.
You cannot add them together to get a total for the park, because the report does not say how the overlap is handled, and there is no "Hirosaki Park" line item at all.
Crucially, this split does not always run the same way. Fukui Prefecture lists Maruoka Castle at 203,000 visitor-days and the Maruoka Castle Cherry Festival separately at 50,000 — the festival is the smaller number there. So "the festival is the big number" is not a rule you can apply.
Hirosaki's cherry-blossom festival is counted separately from the castle's paid area — and is 6.9 times larger. Photo: 掬茶 via Wikimedia Commons, CC BY-SA 4.0.
3. The area versus the building
Okayama Prefecture's 2024 report contains both "Kōrakuen and Okayama Castle area" at 1,824,000 visitor-days and "Okayama Castle" at 423,441 paid admissions — a 4.3× difference between a district and a ticket gate. (Okayama Castle is a 1966 reconstruction; the original keep in that prefecture's table is Bitchū-Matsuyama, at 56,266.)
4. Two rows, same castle, same classification
The clearest case is Matsue. Shimane Prefecture's 2024 survey lists, in adjacent rows of the same table and under the identical classification code for "castle":
- Matsue Castle — 402,049
- Matsue Jōzan Park, the castle park — 452,304
Both are correct. The prefecture's unit is visitor-points, defined in the report as a simple sum across sites in which one traveler visiting several places is deliberately counted more than once. Anyone who climbs the keep has also walked the park.
But the prefecture then publishes a "castle" subtotal of 854,353 — which is exactly 402,049 + 452,304. So if you quote "854,353 people visited castles in Shimane," you have counted one castle twice. The prefecture defined its terms clearly and did nothing wrong. The error, if it happens, happens at the moment of citation.
Someone is already choosing which number gets published
This is the part that changes how you should read any castle visitor figure you see in English.
Kōchi Prefecture's 2024 report includes a reference table of major sites in the three neighboring Shikoku prefectures. For Ehime, the site it chose to represent is the "Matsuyama Castle ropeway and lift," at 1,333,384 (2024) — not the castle keep, which Matsuyama City recorded at 497,511 in the same year. Both numbers appear in primary sources; a public body picked the one that is 2.7× larger.
Kōchi labeled its row accurately, so this is not an error. It is something more useful to know: when a castle has several correct numbers, whoever is quoting gets to pick. By the time a figure reaches an English travel article, that choice has usually already been made and the label has usually been dropped.
Two more cases where the source itself tells you the measuring stick moved:
- Fukui Prefecture states plainly that it updated the conversion coefficients used in its 2024 survey, and prints the before-and-after values (for visitors from within the prefecture: average sites visited per traveler, 1.583 → 1.511; average nights, 1.170 → 1.011). Head-counts are derived by dividing visitor-days by those coefficients, so the same raw footfall produces a different published figure after the update.
- Himeji City changed its counting method in June 2025, moving to records from ticket-machine language selection, digital tickets and counter sales. Himeji publishes a domestic/overseas split — 35.8% overseas in FY2024, 34.9% in FY2025 — but that apparent dip crosses the method change, so it should not be read as a real shift in who is visiting.
And prefectures do not even build their "top sites" tables the same way. Aichi omits sites that did not consent to publication. Shiga includes only sites that agreed. Okayama applies a size threshold — roughly 10,000 visitors a year, or 5,000 in a single month — so a small castle can be absent simply for being small, not for being empty. Shimane goes the other way and keeps sites that have fallen below the threshold. Four prefectures, four rules. "Ranked nth in the prefecture" means "nth among the sites that this prefecture's rules put in the table."
For the same reason, a rank can move without the castle moving. Kōchi Castle recorded 272,037 in 2024, essentially unchanged from 274,652 the year before, while the two sites ranked above it swung by −28.8% and +72.3%. Its third place is a fact about its neighbors.
Before you go: two keeps you cannot simply walk into, and "visitors" doesn't mean people
Three practical points that get lost in the data.
Maruoka Castle's keep is under restoration, and interior access opens and closes in blocks. The castle is in the middle of a major repair project, and the operator publishes the access windows one stretch at a time. Its own notices set out the current schedule: the interior was open through 31 August 2026, is closed from 1 to 18 September 2026 while work is done near the entrance, and reopens on 19 September, staying open through 31 October 2026. Beyond the end of October the operator has not yet published a window. Here is the trap — the admission prices and opening hours stay posted on the site the whole time, including during the closed stretch. A published fee is not evidence that a building is open on the day you arrive. If Maruoka is on your itinerary, read the operator's notices page for the window that covers your dates; do not rely on aggregators, or on this page.
Hirosaki Castle's keep is closed inside altogether — and it is currently moving. This is the easier one to miss, because Hirosaki is one of Japan's best-known cherry-blossom destinations and the park around the keep is fully open and selling tickets. Interior viewing ended on 23 November 2025, and the operator states the keep stays closed until the repair project finishes, currently planned for fiscal 2032 — about seven years. The keep was lifted off its stone base in 2015 for a century-scale stone-wall repair; the stones were restacked by December 2024, and the move back began on 1 July 2026 with a first stage of roughly 19 meters east over four days followed by a 25-degree rotation. A second stage ran through August, a third is scheduled for October, and the keep is due back on its original base in November 2026, along a route the city puts at about 78 meters in total. So the paid area is open, the ¥320 ticket is on sale, and the keep is in plain view — but you cannot go inside it, and it is not standing where the postcards show it.
Most of these figures are not head-counts. "Visitor-days" and "visitor-points" count the same traveler once per site visited, by design. The gap is large and it varies by prefecture: Shimane reports 29,860 thousand visitor-points against 12,451 thousand actual visitors (2.40×), Aomori 2.34×, Fukui 1.84×, Okayama 1.70×. The unit names differ too — Shimane uses nin-chiten (person-points), Kagawa uses nin-kai and nin-paku. You can compare visitor-days with visitor-days, or head-counts with head-counts, but never one against the other.
Bitchū-Matsuyama Castle, at 56,266 paid admissions in 2024, is the least-visited of the twelve — and one of only five figures that come from an actual ticket count. Photo: Reggaeman via Wikimedia Commons, CC BY-SA 3.0.
What we could not establish
Three things we deliberately leave blank rather than fill in.
How many castles Japan has in total. English sources commonly say "about 5,000 castle sites"; Japanese sources commonly say around 25,000. That is a five-fold disagreement, and we could not trace either figure to a primary source. We are not going to pick one. The counts we do publish — twelve surviving keeps, five National Treasures, seven Important Cultural Properties — rest on the national heritage register.
Whether Maruoka is Japan's oldest keep. You will see this claim often. Maruoka's own operator does not make it. Its site calls the castle "the only surviving keep in the Hokuriku region" and says it is "said to have been built in 1576" — a hedge, not an assertion. A scientific dating study was carried out with the prefecture and the city, but its conclusion is not published on the operator's site. So the honest statement is: the national database records 1576, and the people who run the castle describe that date as tradition and do not claim the record.
Which castle is Japan's most visited, full stop. Not from this data. The units and years do not line up, and any answer would require choosing a turnstile — which is precisely the choice this page is about.
There is a reason no comprehensive English-language comparison of Japanese castle visitor numbers exists. It is not that the data is secret; every figure here is published. It is that the data lives in twelve separate Japanese-language reports that were never designed to be compared, and assembling them into a clean ranking would require pretending they measure the same thing.
Sources & methodology
The Japan castle statistics on this page were each retrieved in August 2026 from the publishing body's own report. Search snippets, English guides and aggregator pages were not used as sources.
- Cultural property designations — Agency for Cultural Affairs national heritage database (designation names, dates and periods as recorded).
- Himeji — City of Himeji, castle admission statistics (fiscal year; note the June 2025 method change).
- Inuyama — Aichi Prefecture, tourism and recreation user statistics.
- Matsumoto — Nagano Prefecture tourism statistics (visitor-days).
- Hikone — Shiga Prefecture tourist arrival statistics.
- Iyo-Matsuyama — Matsuyama City, 2025 estimated visitor tables (facility-level monthly figures).
- Kōchi, and the Shikoku reference table — Kōchi Prefecture, 2024 visitor survey report, chapter 2.
- Hirosaki — Aomori Prefecture, 2024 tourism visitor statistics (site-level and event tables); keep closure and move-back schedule from Hirosaki Park's official notice and Hirosaki City's stone-wall repair pages (checked 2 September 2026).
- Uwajima — Ehime Prefecture visitor statistics.
- Marugame — Kagawa Prefecture, 2024 visitor movement survey (final).
- Maruoka — Fukui Prefecture, 2024 estimated visitor numbers; keep access windows from the castle's official notices (re-checked 2 September 2026).
- Bitchū-Matsuyama and Okayama — Okayama Prefecture, 2024 visitor movement survey report.
- Matsue — Shimane Prefecture, 2024 tourism survey results.
Units. Paid admissions / ticketed = counted at a gate, one per entry. Visitor-days (延べ) and visitor-points (人地点) = summed across sites, so one traveler may appear several times. Estimate = derived from transport or lodging data using published coefficients.
Update policy. Designation facts are stable and are not expected to change. Visitor figures and admission fees are refreshed annually as each publishing body releases its next edition; the year is shown against every figure so you can see how current it is.
Image credits: Cover image — Matsumoto Castle and the Hida Mountains, Nagano Prefecture, by 663highland (CC BY-SA 4.0). In-text images are credited in their captions. All images via Wikimedia Commons.
Frequently Asked Questions
Which Japanese castle gets the most visitors? There is no single answer, and the reason is in the source data. The largest ticketed count among the twelve original keeps is Himeji, at 1,567,674 paid admissions in FY2025. But Iyo-Matsuyama records 1,638,438 if you count ropeway and chairlift passengers, and 612,203 if you count keep tickets. Different turnstiles, different answers, all correct.
How many original castle keeps are left in Japan? Twelve: Hirosaki, Matsumoto, Maruoka, Inuyama, Hikone, Himeji, Matsue, Bitchū-Matsuyama, Marugame, Iyo-Matsuyama, Uwajima and Kōchi. Everything else standing today is a reconstruction or a restoration.
Which Japanese castles are National Treasures? Five: Matsumoto, Inuyama, Hikone, Himeji and Matsue. Matsue is the most recent addition, designated in July 2015 — sixty-three years after the previous three.
Can I compare castle visitor numbers between prefectures? No. The years differ, the counting units differ, and the rules for which sites appear in a prefecture's table differ from prefecture to prefecture. Only five of the twelve figures here are actual ticketed counts. Compare like with like or not at all.
Are all twelve original keeps open to visitors? No — two of them are shut inside right now, for different reasons. Hirosaki's keep stopped admitting visitors inside on 23 November 2025 and stays closed until its repair project finishes, currently planned for fiscal 2032; the keep itself is being hauled back onto its original stone base in stages through November 2026, so it is not even standing in its usual spot. Maruoka Castle is under a major restoration too, and its own website publishes interior access one window at a time: it was open through 31 August 2026, is closed 1–18 September 2026, and reopens on 19 September, running through 31 October 2026, with no window published beyond that yet. Maruoka's admission prices and opening hours stay posted throughout, including during the closed stretch, so check the operator's notices page for your travel dates rather than assuming a posted fee means open doors.
Final Thoughts
If you came here for a league table of Japan's castles, the useful finding is that a trustworthy one cannot be built from the available data — and that this is not a gap in Japanese record-keeping. The prefectures and cities publishing these numbers define their terms carefully, print their coefficients, and flag it when their method changes. Everything falls apart one step later, at the moment somebody lifts a single figure out of its table and drops the label.
Which leaves a better question than "which is the biggest?" Twelve keeps came through four centuries, and one of them draws fewer visitors in a year than Himeji draws in a fortnight. Bitchū-Matsuyama sits at the top of a mountain in Okayama and counted 56,266 people in 2024. That is not a lesser castle. It is a quieter one.
Related reading: Matsumoto Castle · Matsue and its castle town · Hirosaki Castle in cherry-blossom season · Dōgo Onsen and Matsuyama · Japan cherry blossom statistics · Japan overtourism statistics · Tōhoku travel guide