Mutation rates for X-linked recessive diseases have so far been estimated indirectly by postulating an equilibrium between the loss of defective genes caused by the low reproductive fitness of affected males and the gain resulting from new mutations. Here, for the first time, we directly estimate both the overall and sex-specific mutation rates for haemophilia B by detecting the gene defect of the families registered at the Malmö Haemophilia Centre. These represent a complete sample of the Swedish haemophilia B population (45 out of 77 pedigrees) and contain 23 families with a single affected male. Fifteen of these males had mothers available for study, and of these mothers, 13 had parents available for study. We show that 3 of the above patients and 10 of their mothers carry new mutations, and by extrapolation calculate that 8 males and 98 females should carry new haemophilia B mutations in the Swedish population (8.52 x 10(6) individuals). This leads to the following estimate of the mutation rates: overall mu = 4.1 x 10(-6); male specific nu = 2.1 x 10(-5); and female specific mu = 1.9 x 10(-6). The ratio of such male to female specific mutation rates is thus nu/mu = 11.