Knowing how a Japanese survey respondent answered any one question would have improved a guess at their blood type by at most 0.04 percentage points over always guessing A, on our model of the tables Kengo Nawata published in 2014. The belief his study tested, popularised in Japan from 1971, holds that ABO blood type shapes temperament and decides which types get along. Nawata's three surveys in Japan and the US analysed nearly 9,700 people who gave their blood type, and on each of 68 tests blood type explained less than 0.3% of the variation in answers. The rare positive findings elsewhere are faint, or turn up mainly among people who hold the belief.
Large surveys in Japan and the US show almost no difference between the types
Nawata's paper in The Japanese Journal of Psychology used no personality test. It reanalysed an economics survey of attitudes to life and money, run by Ikeda, Ohtake and Tsutsui, from Japan in 2004 and 2005 and the US in 2004. The Japanese samples were random draws from the resident register; the American one was a research firm's postal panel. Respondents reported their own blood type.
The abstract's "over 10,000" matches the 11,729 questionnaires returned; by our sum, the analyses used 9,533 to 9,722 people, depending on the question. About 29 distinct questions, by our count, gave 68 tests, and 65 found no significant difference between the four types. Nawata notes that three or more of 68 would cross the 5% line by chance 66.7% of the time. The one question significant in both countries, worry about one's children's future, pointed opposite ways: type AB scored highest in Japan's 2004 survey and lowest in the US.
Our guessing figure uses only the two Japanese surveys and treats each answer as normally distributed around each type's printed mean, one question at a time. That model is ours, and approximate, since answers sit on a five-point scale. Always guessing A, the largest group among Nawata's respondents, would have been right 38.83% of the time in the 2005 survey; the most informative single question raised that to 38.87%.
Faint links appear in a genotyped sample and among believers
Tsuchimine and colleagues genotyped 1,427 healthy Japanese adults in a 2015 PLoS One study and gave them the Temperament and Character Inventory. They were testing a specific idea, proposed by Hobgood in 2011, that a gene called DBH lies "in tight linkage with the ABO gene". They found a weak association with one trait, Persistence, which averaged 4.6 for type A against 4.3 for B and O. The authors call the effect small and the result one that "should be regarded as preliminary".
Other positive findings lean on belief. Yamaoka (1999, 2006), as Nawata reports, found differences among believers in blood-type typing and almost none among non-believers. Sakamoto and Yamazaki (2004) saw type A respondents in 1978–88 data drift towards the type A stereotype, and read it as a self-fulfilling prophecy; Nawata's 2000s Japanese data showed no such differences. In Korea, Cho, Suh and Roh surveyed 204 college students in 2005. Their abstract reports "no meaningful relationship" between Big Five traits and blood type, while firm believers rated themselves "in a stereotype-consistent direction" and skeptics did not.
Furukawa's own paper reveals a hypothesis tried out on his relatives
The academic version that caught on came from Takeji Furukawa of the Tokyo Higher Normal School for Women, in a 1927 paper in the same journal, 「血液型による氣質の研究」 (J-STAGE). Nawata writes that it drew wide attention and that later replications judged it academically unsound. Furukawa began with 11 of his own blood relatives, then scored 50 colleagues and friends, counting 46 as fitting. He then had 269 students blood-typed, of whom 248 handed in answers. His foreword admits 「不備の點が多い」, many shortcomings.
Nomi's titles of the 1970s show the idea marketed as science
Nawata traces the modern craze to a 1971 book by the writer Masahiko Nomi, 『血液型でわかる相性』 (compatibility you can tell from blood type). By our count of the National Diet Library catalogue, Nomi published a dozen distinct blood-type titles before his death in 1981. Two subtitles from 1979 borrow the vocabulary of the laboratory: 「幸せになる相性の科学」 (the science of compatibility for happiness) and 「一度きりの人生を決める科学的プラン」 (a scientific plan for the one life you get).
Contradictions surfaced in our own compatibility notes
Lore of this kind has nothing outside itself to check against, only its own consistency, and ours failed that test. Our blood type personality tool presents the profiles as entertainment with no scientific basis. Until 7 October 2026 its pair notes and each profile's best-match line were written separately, and 2 of the 10 pair notes disagreed with the best-match line beside them. A fix made that day reads both from one table, labels the profiles as our summary of the lore, and replaces a claim of "tens of thousands" of study participants with Nawata's figures. The live page may show the old version until the fix is deployed.
Japan's broadcasting ethics body brought viewer complaints into the open in 2004
On 8 December 2004 the youth committee of the Broadcasting Ethics & Program Improvement Organization (BPO) issued a request about programmes on blood type. Classifying and valuing people by a blood type they cannot change, it wrote, 「社会的差別に通じる危険がある」: it risks leading to social discrimination. It judged that such programmes risked breaching Article 54 of the commercial broadcasters' code as it then stood, asked broadcasters not to promote the view that blood type determines personality, and objected to children being used as subjects in televised experiments.
The details of harm come from viewer letters the committee published with the request. A parent wrote that bullying had started at their child's school with the words "you must be type B". A teenage viewer called the equation 「性格悪い＝Ｂ型」, bad personality equals type B, hurtful.
Blood-bank tables from Singapore and Malaysia show O well ahead of A
ABO type is inherited, as the Japanese Red Cross sets out, and the mix varies by place. In Japan type A is the largest group, at about 40% by the Red Cross's approximation; Nawata's 2004 Japanese respondents came to 37.8%, by our arithmetic. For Singapore and Malaysia, the two ASEAN members we checked, the published figures count donors and donations, which reflect who comes forward to give blood.
| Place | What the figure measures | O | A | B | AB |
|---|---|---|---|---|---|
| Japan | Approximate share of the population (Japanese Red Cross) | 30% | 40% | 20% | 10% |
| South Korea | Approximate share of the population, undated (Korean Red Cross) | 28% | 34% | 27% | 11% |
| Singapore | 77,567 donors in 2025, Rh groups summed by us (HSA) | 44.7% | 24.4% | 24.9% | 6.0% |
| Malaysia | 589,834 donations in 2025 at 22 health ministry sites, our sum (MOH Malaysia / National Blood Centre, CC BY 4.0) | 42.7% | 24.7% | 27.0% | 5.6% |
The rows measure different things, so they do not rank the four places against one another. Within Singapore's own donor figures, though, A comes third, a fraction behind B, and the bet that served best in Nawata's Japanese samples would lose to always guessing O.
View author profile → · Editorial policy
