A new series of 96 pedigrees with the fra(X) syndrome was analysed using complex segregation analysis with pointers, defining affection as any degree of mental impairment. These families were found to exhibit the same segregation pattern as the first series of 110 pedigrees (Sherman et al. 1984). The best estimate for penetrance of mental impairment in males was 79% and in females was 35% for the combined data. Again, there was little evidence for sporadic cases among affected males. Many more intellectually normal transmitting males have been observed since the existence of such males and the concomitant need to investigate the paternal side of pedigrees was recognized. On further investigation of all 206 pedigrees from the old and new data sets, the sibships of nonexpressing males appeared to be different from those of expressing males. Our analysis, using mental impairment as the phenotype, suggested that obligate carrier mothers and daughters of intellectually normal transmitting males are rarely, if ever, mentally impaired and that the sibs of transmitting males are much less likely to be retarded than the sibs of mentally impaired males. Though mothers and daughters of transmitting males are similar in phenotype, the expression of the gene in their offspring appears to be different: the penetrance of mental impairment is higher in offspring of intellectually normal daughters of transmitting males than in offspring of intellectually normal mothers of transmitting males. The implications of these observations for genetic counseling and for genetic models of the fra(X) syndrome are discussed.