Under the textbook ABO rule, every pairing of parent types except A with B forbids at least one child type, 16 of 40 pairing-and-child combinations by our count. In our genotype model, rare variants reach 11 through a parent carrying the Bombay phenotype or cis-AB, one more through a mistyped parent, and another in one Korean family with a new mutation; A with A giving B, and B with B giving A, have no route in any source we opened. Where the variants cluster most, as cis-AB does in Korea, they reach a few children in a thousand, a population average that explains no single family's result. In the papers we read, they surfaced through typing discrepancies or family studies.
The textbook rule forbids a child type to every parent pairing except A with B
A blood group on a card is a phenotype, read off the red cells; behind it sit two ABO alleles (gene versions), one from each parent. In the textbook, A and B are codominant and O recessive; RhD-negative means two d alleles.
Our enumeration of all 64 ordered ABO and RhD parent combinations gives, for ABO alone:
| Parent types | Forbidden child types | Route in our model |
|---|---|---|
| A and A | B, AB | AB via a mistyped parent |
| A and B | none | none needed |
| A and AB | O | Bombay; cis-AB |
| A and O | B, AB | Bombay |
| B and B | A, AB | AB in one new-mutation family |
| B and AB | O | Bombay; cis-AB |
| B and O | A, AB | Bombay |
| AB and AB | O | Bombay; cis-AB |
| AB and O | O, AB | Bombay; cis-AB |
| O and O | A, B, AB | A or B via Bombay |
O with O giving AB needs one parent carrying both variants, which no source we opened reports. For RhD the rule forbids two RhD-negative parents an RhD-positive child; our sources document no exception.
Cis-AB allows an AB parent to have an O child
The ISBT allele table of 23 October 2017 (209 ABO alleles, by our count) says some make "promiscuous enzymes that can synthesize both A and B antigen". One, cis-AB, carries A and B together, so a parent with cis-AB and O types as AB yet passes on O half the time. Related B(A) alleles go unmodelled here.
Cis-AB is most often reported in Korea. Chun and colleagues' 2019 review cites 60 cases in 169,605 Korean donors (Cho and colleagues, 2004), a lower bound since only unusual AB samples were screened. With the Korean Red Cross figure of about 11% typed AB, we find about 1 in 311 AB-typed Koreans carry it. About 1 in 389 children of Korean AB and O parents then fall outside the textbook A or B.
Those averages, from a 2004 donor series and an undated population share, cannot explain any one family's result: a parent with cis-AB and O passes an out-of-list type to every child, a non-carrier, by this route, to none.
The review also describes four Korean people with cis-AB and A who "would have been typed as typical A" without careful family studies; through such a parent, A with A gives AB. Its single family of two group-B parents whose child carried a new cis-AB mutation opens no general route.
Bombay-phenotype parents permit A or B children while typing as O
Bhende and colleagues described the phenotype in 1952 in The Lancet; Bombay is Mumbai's former name. People with it lack the H antigen on which A and B are built: "ABH antigens are absent from their red cells, regardless of their ABO blood group genotype" (Shahshahani and colleagues, 2013). They type as O yet pass on A or B genes, reaching 11 forbidden combinations in our model, though no case report we opened describes such a family.
An earlier study cited by Joshi and Vasantha (2012) put the phenotype at 1 in 7,600 in Mumbai. Borrowing an O share from another Indian hospital study, we bound the chance of a non-O child of two O-typed parents at under 1 in 1,000 in Mumbai. Such population figures explain no one family.
Donor shares let each permitted child type have its own odds
Singapore's Health Sciences Authority (HSA) lists 2025 donor shares by type (updated 31 May 2026), tabulated in our earlier guide on blood types. Under Hardy-Weinberg proportions, which assume parents pair without regard to blood type, an A-typed donor carries a hidden O about 89% of the time, so two A parents have an O child about 1 in 5 times, on our arithmetic.
For RhD, HSA's 2025 negative types sum to 1.3% of donors, giving two RhD-positive parents an RhD-negative child about 1 in 100 times. HSA's Rhesus page (updated 13 March 2026) says "about 1.5%", and a release of 22 July 2009 put negative types at "less than one per cent of the Singapore donor population" and added: "These blood types are more commonly found in the Caucasian and Indian communities." The Korean Red Cross's 0.1% of Koreans gives roughly 1 in 1,000.
These illustrate how much "possible" hides and predict nothing. Donors are not the population, Singapore is multi-ethnic so a pooled share misstates every family, and Hardy-Weinberg assumes random pairing.
Our blood type inheritance calculator lists permitted types without odds, and its textbook logic matched our enumeration. Until 8 October 2026 it called its outcomes certain, stated that AB and O parents could not have an O child, and treated an excluded type as a conclusion about a family, which was wrong. The revision made that day labels the grid a textbook model and names documented routes.
RhD-negative donations under Malaysia's guidelines require a weak-D test
The National Blood Centre's Transfusion Practice Guidelines (fourth edition, June 2016, the latest we found) require that "all donated blood that is typed as RhD negative shall be subjected to a weak-D test". Weak D and DEL carry D antigen too faint for a first test; DEL "types as RhD-negative in routine serology" (Yin and Flegel).
Their 2021 review compiles 16 studies of serologic RhD-negative Han individuals, with DEL at 16.3–32.6% across them and 23.3% pooled. In a Japanese Red Cross donor study (Ito and colleagues, 2021), 240 of 2,754 serologic RhD-negative donors were DEL, 8.7% by our arithmetic. No source we opened says what DEL means for a child's predicted type or for care.
Reverse grouping against O cells lets a laboratory detect a Bombay sample
Forward grouping tests the red cells; reverse grouping tests the serum against cells of known type. Malaysia's guideline runs reverse grouping with A1, B and O cells and says "the O-cell test is incorporated to detect Bombay blood group".
Shastry and colleagues (2013) describe a woman "wrongly typed as O positive at the peripheral center", an error that "would not have happened" with serum grouping against pooled A, B and O cells. Shahshahani's patient was "misdiagnosed as the O blood group". Chun's review, citing an earlier study, reports cis-AB samples misread as typical AB by a microplate device.
At the donation site, the guideline adds, a finger-prick group "should only be considered a preliminary result of the blood group and should not be used for any other purpose whatsoever."
View author profile → · Editorial policy
