The results of a cytogenetic and segregation analysis of 110 pedigrees of the mar (X) syndrome are reported. The cytogenetic study indicated an inverse relationship between IQ and the mar(X) frequency in females but not in males. A small but significant effect of age on mar(X) frequency was observed in both males and females, but in females it was restricted to those of normal intelligence, retarded females showing no significant effect. Classical segregation analysis was performed using the program SEGRAN, analysing sexes separately. A 20% deficit of affected males was observed, the most plausible explanation for the majority of these cases being incomplete penetrance. Since this was an unexpected result, the data were scrutinized for possible biases; however, correction of these had little effect on the estimate. The penetrance of mental impairment in carrier females was estimated to be 30% and of mental impairment and/or mar(X) expression to be 56%. Thus 44% of carriers cannot be detected with our definition of affection. No evidence for sporadic cases among affected males was found. Complex segregation analysis was performed using the sex-linked version of POINTER, analysing sexes together. This was done in order to test the results from classical segregation analysis, to test for family resemblance and to estimate mutation rates. It was confirmed that there was a 20% deficit of affected males, that penetrance of mental impairment in females was approximately 30% and that there was no evidence for sporadic males. Thus all males with the gene appear to have received it from their carrier mothers and all mutations must occur in sperm. The mutation rate in sperm was estimated to be as high as 7.2 X 10(-4), implying that over one-half of random carrier females are fresh mutants. Our results have important implications for genetic counseling as they imply that all mothers of isolated affected males are carriers, that normal brothers of affected males have a 17% chance of carrying the gene and transmitting it to all their daughters, and that normal sisters of affected males have, at most, a 30% chance of being carriers. Since there are biases in the data due to the testing of particular individuals, these probabilities must be considered approximations until they are independently confirmed.