In winter 2027, the Kyoto Marathon (Sunday, February 21), the Osaka Marathon (Sunday, February 28) and the Tokyo Marathon (Sunday, March 7) fall on three consecutive Sundays. Field sizes are 16,000 in Kyoto, 34,000 in Osaka and 40,000 in Tokyo — a combined 90,000 runners moving from Kansai to Kanto within three weeks. Looking back at the eve of the previous 2026 races through listed prices on major booking sites, Kyoto rose +28% over a normal Saturday around Heian Shrine and Tokyo +14% around the Tokyo Metropolitan Government Building, while the Osaka Castle area posted the largest jump at +59% — although that night coincided with the first day of the Emperor’s Birthday three-day weekend. This article lines up the dates, field sizes and day-of-week placement of all three races in one place, then examines the price response on the eve and the night of the previous races, monthly occupancy from Japan Tourism Agency statistics, the field sizes converted into pre-race room-night demand, and the room stock around each start and finish.

Race Dates and Field Sizes — Three Consecutive Sundays in 2027, None Overlapping a Holiday

First, the dates for this edition. The Kyoto Marathon 2027 takes place on Sunday, February 21, 2027, starting at Takebishi Stadium Kyoto (Nishikyogoku Athletic Park) and finishing in front of Heian Shrine. The field is 16,000 for the marathon (including organizer-recommended entries) plus 250 pairs (500 runners) in the pair ekiden, the same scale as the previous race. The Osaka Marathon 2027 starts in front of the Osaka Prefectural Government Office on Sunday, February 28 and finishes inside Osaka Castle Park. The entry guidelines set a total field of 34,000 — 31,970 for the marathon and 2,030 for the 720 “Naniwa” Marathon — also unchanged from the previous edition. The Tokyo Marathon 2027 runs on Sunday, March 7, departing from the Tokyo Metropolitan Government Building and finishing at Gyoko-dori in front of Tokyo Station. Its field is set at 40,000, an increase of 1,000 from the previous 39,000 (38,500 marathon and 500 for the 10.7km).

Comparing the day-of-week placement with the previous year, the difference lies in Osaka. In 2026, the eve of the Osaka Marathon, Saturday February 21, coincided with the first day of the three-day weekend for the Emperor’s Birthday (Monday, February 23). In 2027, February 23 falls on a Tuesday, so the Osaka race weekend (February 27–28) is separated from the holiday. In Kyoto, as in the previous year, the race does not sit next to a holiday in 2027, but taking Monday, February 22 off after the race connects to the holiday on Tuesday the 23rd, creating room to extend stays from race night into the following night. In Tokyo, neither 2026 nor 2027 overlaps a holiday. In other words, 2027 is a year in which “marathon demand itself” creates the race-eve peak for all three events.

Source: Compiled by HotelBank Editorial Team from the Kyoto Marathon 2027 event guidelines, Osaka Marathon 2027 entry guidelines and Tokyo Marathon 2027 event outline (all checked October 27, 2026)

Eve and Night of the Previous (2026) Races — Demand Peaks Seen Through Listed-Price Response

Daily listed prices are the most direct reflection of where demand peaked. The chart below plots, for the area around each race’s start and finish (Tokyo: 1km radius of the Metropolitan Government Building; Osaka: 2km radius of Osaka Castle Park and the Prefectural Government Office; Kyoto: 2km radius of Heian Shrine), the property median of listed prices from January 31 to March 14, 2026, indexed to each area’s all-day February median = 100. The number of properties covered each day was 18–26 in the Metropolitan Government Building area, 69–84 in the Osaka Castle area and 95–119 in the Heian Shrine area.

Source: MetroEngines Research; compiled by HotelBank Editorial Team

On the Kyoto race eve, Saturday February 14, the Heian Shrine area median was ¥36,000, +28% over the surrounding normal Saturdays (January 31 ¥28,100, February 7 ¥28,100, February 28 ¥25,400). Race night, Sunday February 15, was also ¥23,800, +19% above other February Sundays (¥19,900–¥20,000), with prices reflecting runners staying in Kyoto that night after finishing. Meanwhile, Saturday February 21, with no race, rose to ¥41,300 as the first day of the three-day weekend, showing that in Kyoto the holiday-calendar effect exceeded the marathon-eve effect.

On the Osaka race eve, Saturday February 21, the Osaka Castle area median reached ¥28,700, +59% over normal Saturdays (¥17,400–¥18,400) — the largest lift of the three areas. However, that day was also the first day of a three-day weekend. Given that on the same Saturday, without a race, Kyoto rose +47% and the Tokyo Metropolitan Government Building area +12%, Osaka’s +59% is best read as the combined result of the three-day weekend and the marathon. Race night, Sunday February 22, was ¥19,100, +43% above normal Sundays (¥12,700–¥13,800), but this too includes the mid-holiday effect. Because Osaka in 2027 is separated from the holiday, the height of the eve peak will be determined by the race’s own drawing power. Rather than assuming the previous +59%, it is more realistic to use Kyoto’s +28% as a reference point.

In Tokyo, the Metropolitan Government Building area median on the eve, Saturday February 28, was ¥50,600, +14% over the median of normal Saturdays (January 31 ¥42,900, February 7 ¥42,500, February 14 ¥47,200, March 7 ¥45,500) of ¥44,200. The Tokyo Station area on the finish side was ¥57,000, below February 21, the first day of the three-day weekend (¥67,600), and the following week’s March 7 (¥63,000); around Tokyo Station, the race-eve response barely shows in prices. This is likely because the baseline Saturday level is already high and race demand “fills in” Saturdays when business demand is thin. On race night, Sunday March 1, the Metropolitan Government Building area was ¥35,800, +27% over February Sundays (¥26,500–¥29,000), and the Tokyo Station area at ¥51,500 also exceeded early-February Sundays. Tokyo differs from Kyoto and Osaka in that the relative lift is larger on race night than on the eve.

Source: MetroEngines Research; compiled by HotelBank Editorial Team (listed price = property median of each property’s all-plan average, 2 guests per room, tax included. “Normal Saturday” is the median of Saturdays from January 31 to March 7, excluding race eves and the first day of the three-day weekend)

Monthly View: Occupancy Bottoms in January, Rates in February — Japan Tourism Agency Statistics and Ward-Level ADR

So where does the January–March period in which the three races sit fall on a monthly basis? Lining up guest-room occupancy from the Japan Tourism Agency’s Overnight Travel Statistics Survey (second preliminary figures) for January–March 2026, Tokyo was 70.7% in January, 76.8% in February and 73.6% in March; Osaka 62.8%, 70.1% and 70.5%; and Kyoto 53.3%, 59.1% and 64.4%. All three bottomed in January; in February Tokyo returned to one of its higher levels of the year, while Kyoto climbed step by step toward March. Osaka was down 6.3pt and 7.5pt YoY in February and March respectively, falling below the same months of the previous year.

Source: Compiled by HotelBank Editorial Team from the Japan Tourism Agency Overnight Travel Statistics Survey (January, February and March 2026, second preliminary figures)

Looking at rates through ward-level estimated settled ADR (this article’s ADR, monthly property median), the bottom is slightly offset from occupancy. Osaka’s Chuo Ward was ¥11,300 in January 2026, ¥10,300 in February and ¥11,400 in March, bottoming in February, with February down 12.1% YoY. Kita Ward shows the same shape at ¥10,400, ¥10,100 and ¥10,800. Kyoto’s Shimogyo Ward was ¥11,400, ¥10,900 and ¥18,900, bottoming in February before jumping sharply in March. Nakagyo Ward was ¥12,500, ¥12,600 and ¥21,000, with February at roughly the same level as January and a jump in March. In Tokyo, Shinjuku Ward was ¥15,500, ¥15,300 and ¥17,200, and Chiyoda Ward ¥15,000, ¥15,000 and ¥17,900, with January and February level and a rise in March. In short, February is the ADR trough in Osaka and Kyoto, and within that trough a one-day peak stands on the marathon eve. This is why the race-eve lift stands out within monthly ADR.

Source: MetroEngines Research; compiled by HotelBank Editorial Team (estimated settled ADR, monthly property median; N varies by month)

Converting Field Sizes into Pre-Race Room Demand — 17,000–24,000 Rooms Across the Three Races

To estimate the size of race-eve demand in rooms, we scale the field size by the share of runners from far away, the pre-race stay rate and the number of guests per room. The basis for the far-away share is the Kyoto Marathon 2026 runner survey (N=1,973): residence was 23.5% Kyoto City, 6.4% elsewhere in Kyoto Prefecture, 13.5% Osaka, 6.7% Hyogo, 3.3% Shiga, 1.9% Nara, 0.5% Wakayama, 8.4% Tokyo, 5.5% Aichi, 4.7% Kanagawa, and so on. Runners from outside Kyoto Prefecture reached 69.6%, but those from other Kinki prefectures (25.9%) live within day-trip distance. Limiting to outside the Kinki region gives 43.7%, which we use as the share of “runners for whom a pre-race stay is realistic.” Comparable published figures could not be confirmed for Osaka and Tokyo, so we assumed a range of 35–50% based on Kyoto’s structure. A pre-race stay rate of 80% of far-away runners and 1.5 guests per room (including companions) are common assumptions.

Source: Compiled by HotelBank Editorial Team from field sizes on each race’s official website and the Kyoto Marathon 2026 runner survey results (Kyoto Marathon Executive Committee, N=1,973). The far-away share, pre-race stay rate and guests per room are this article’s assumptions

The converted values are only a stack of assumptions and exclude companions, supporting family members and multi-night stays by overseas runners. Kyoto has an overseas runner quota of 3,000 (2027; 2,031 entries as of end of August), Osaka has 200 overseas charity entries plus overseas participants in its general category, and Tokyo cites growth in overseas participation as a reason for revising its entry fee. Overseas participants do not stop at one night on the eve but stay multiple nights, so it is natural to expect actual demand to be on the heavier side of the table above.

The two inputs that move this conversion the most are the far-away share and guests per room. The table below shows the three-race total when both assumptions are varied simultaneously, with the pre-race stay rate fixed at 80%. Within a far-away share of 35–50% and 1.3–1.8 guests per room, pre-race demand ranges from 14,000 to 28,000 rooms; the central case — applying Kyoto’s survey actual (43.7% from outside Kinki) to all three races with 1.5 guests per room — gives approximately 21,000 rooms. The more companions (and thus guests per room), the smaller the room count; the more runners travel solo, the larger it swings.

Source: Compiled by HotelBank Editorial Team from field sizes on each race’s official website (three-race total 90,000) and the Kyoto Marathon 2026 runner survey results (Kyoto Marathon Executive Committee, N=1,973). The far-away share, pre-race stay rate and guests per room are this article’s assumptions; each cell is calculated directly from the total field and rounded to the nearest 1,000 rooms

Room Stock Around the Starts and Finishes — Pre-Race Demand Equals 40–70% of Nearby Rooms

To see how many rooms are available to absorb the converted pre-race demand, we counted rooms within a radius of each start and finish. Room stock covers lodging facilities with 10 or more rooms (excluding leisure/love hotels); “active” refers to properties with selling prices confirmed on major booking sites in the most recent 3 months (August–October 2026) (as of October 28, 2026, compiled by MetroEngines Research).

Source: MetroEngines Research; compiled by HotelBank Editorial Team

Source: MetroEngines Research; compiled by HotelBank Editorial Team (room stock as of October 28, 2026. “Listed” = properties listed on major booking sites; “active” = properties with selling prices confirmed in the most recent 3 months)

Comparing converted pre-race demand with active rooms within each radius: in Tokyo, 7,500–10,700 rooms against a combined 15,600 rooms in the Metropolitan Government Building and Tokyo Station areas, equivalent to 48–69% of nearby stock. In Osaka, 6,300–9,100 rooms against 15,000 rooms within 2km of Osaka Castle, or 42–61%; in Kyoto, 3,000–4,300 rooms against 8,400 rooms within 2km of Heian Shrine, or 36–51%. In every case the demand is “not something the start/finish areas alone can absorb, but easily absorbed when spread across the whole city center (5km radius: Tokyo 103,000 rooms, Osaka 85,000, Kyoto 51,000).” This is why the race-eve peak spreads thinly across the city center rather than concentrating in a specific area. Results of measuring rooms within 5km of the start for 12 autumn-to-winter races using the same approach are summarized in 12 Marathons Oct–Feb: Start-Area Rooms Measured.

Kyoto alone has a different structure. Within 2km of the start at Nishikyogoku Athletic Park there are just 3 active properties with 46 rooms — effectively a blank — so runners travel to the venue from central Kyoto by subway, Hankyu or race shuttle. Pre-race accommodation splits between Shimogyo and Nakagyo wards (around Kyoto Station and Shijo) and Sakyo and Higashiyama wards near the finish. While the race-eve price response showed clearly at +28% in the Heian Shrine area, race-eve demand itself should be understood as dispersed across the whole city.

Implications for Operators — Pricing and Inventory Built Around Day-of-Week Placement

Lining up the three races shows that the height of the race-eve peak depends more on “day-of-week placement” than on “race size.” Osaka, the highest last time, benefited from overlapping the first day of a three-day weekend, a tailwind absent in 2027. Conversely, Kyoto posted +28% last time even without an adjacent holiday, and in 2027 the holiday two days later adds extended stays from race night into the following night. Tokyo’s eve lift is moderate — +14% in the Metropolitan Government Building area and roughly flat around Tokyo Station — but it has a distinctive pattern in which race night exceeds February Sundays by +27%.

Organizing the operational points from this: first, for the Osaka eve (Saturday, February 27, 2027), rather than expecting the previous price level, setting a baseline of +20–30% similar to Kyoto and deciding the size of increases while watching bookings build is less likely to leave revenue on the table. Next, Kyoto’s race night (Sunday, February 21) and the following night (Monday, the 22nd) form a rare combination of weekdays before a holiday, leaving considerable room to combine multi-night plans and late check-out the next day. For Tokyo, given the tendency for a relatively larger upswing on race night (Sunday, March 7) than on the eve, pricing Sunday night separately from “early-arriving business demand for the start of the week” is an opportunity. How to spread demand concentrated on the eve to race night and the night before the eve is also covered in our Shimonoseki Kaikyo Marathon booking curve analysis.

In every city, the relevant 2027 dates were more than 45 days out from the stay date at the time of writing. It is too early to declare sell-outs or tightness from current inventory, and this article makes no such judgment. Checking booking progress from 45 days before each race (early January for Kyoto, mid-January for Osaka, late January for Tokyo) is sufficient; until then, it is enough to pin down only “where the peaks will fall” from the previous price response and this year’s day-of-week placement. For an example comparing pre-race demand across multiple cities at the same 45-day lead time, see our pre-race analysis of the Kanazawa, Yokohama and Kobe Marathons 2026.

Summary

The three big-city marathons of winter 2027 line up on three consecutive Sundays — Kyoto February 21, Osaka February 28, Tokyo March 7 — with a combined field of 90,000, which converts into pre-race demand of 16,800–24,100 rooms within the assumption range. On the eve of the previous 2026 races, listed prices rose +28% in the Heian Shrine area, +14% in the Metropolitan Government Building area and +59% in the Osaka Castle area (overlapping the first day of a three-day weekend), and peaks persisted on race night as well, at +19% in Kyoto and +27% in Tokyo’s Metropolitan Government Building area. On a monthly basis, occupancy bottoms in January and ADR bottoms in February in Osaka and Kyoto; the marathon eve is a one-day peak standing within that trough. Active rooms around the starts and finishes total 16,000 within 1km in Tokyo, 15,000 within 2km in Osaka and 8,400 within 2km of Heian Shrine in Kyoto; pre-race demand equals 40–70% of that, so demand spreads across each city center. In 2027 none of the three races overlaps a holiday, making it a year in which the height of the race eve is determined by each race’s own drawing power.

Frequently Asked Questions

Q. What are the dates and field sizes of the three 2027 races?
A. The Kyoto Marathon 2027 is on Sunday, February 21, 2027, with a marathon field of 16,000; the Osaka Marathon 2027 is on Sunday, February 28, with 34,000 in total (31,970 marathon); and the Tokyo Marathon 2027 is on Sunday, March 7, with a field of 40,000 (each race’s official website, checked October 27, 2026).

Q. How much did hotel prices rise on the eve of the previous races?
A. For the eve of the 2026 races, based on the property median of listed prices around the starts and finishes, the Kyoto Heian Shrine 2km area was +28% over a normal Saturday, the Tokyo Metropolitan Government Building 1km area +14%, and the Osaka Castle Park 2km area +59%. However, the Osaka eve coincided with the first day of the Emperor’s Birthday three-day weekend, and the figure includes that effect.

Q. How is pre-race room demand calculated?
A. Field size × far-away share × pre-race stay rate of 80% ÷ 1.5 guests per room. The far-away share is based on the Kyoto Marathon 2026 runner survey (N=1,973), in which 43.7% came from outside the Kinki region; for Osaka and Tokyo we assumed a range of 35–50%. Companions and multi-night stays by overseas runners are not included.