Usually, the rules of inheritance are fairly straightforward. The A and B genes are codominant (i.e. if both genes are present, neither overrides the other), but the O gene is best thought of as recessive. Therefore:
- Type A people can have AA or AO genes (alleles) - either one will end up displaying only A antigens.
- Similarly the type B people can have BB or BO genes.
- People with blood type AB can only have AB as alleles.
- Type O people can only have OO as alleles.
In terms of the Rhesus group (i.e. + or -), this is dominantly expressed. This means that the only way for a child to be rhesus negative is for it to receive two recessive alleles - one from each parent. Therefore, it is possible for two Rhesus positive parents to produce a Rhesus negative child, but only if they are both (a) have recessive alleles and (b) pass them on.
As complicated as this sounds, if this were the whole picture, then the possible blood types can be worked out by simple mendelian inheritance rules. I've worked all the possible permutations out at the end of this post.
However, this simplistic picture doesn't always hold true. You can actually produce blood type O from parents who are A and AB. It is fairly rare but extremely well documented. For one of the most common ways that this occurs, read about the Bombay Phenotype.
Rhesus studies too are not reliable enough to determine parentage. Again, the simple mendelian rules are usually correct, but not always. For instance, take a look at the 'weak D phenomenon'.
Establishing parentage by the ABO and Rhesus systems has long fallen out of favor amongst those people who know what they are talking about. At best, it is suggestive, but hardly conclusive. Parentage should be established through DNA analysis, and not by other inferior means.