X-linked mental retardation with fragile X or Martin-Bell Syndrome (MBS) is a frequent cause of mental retardation. So far segregation analysis of MBS in pedigrees ascertained by different, incomplete criteria has produced results, difficult to interpret, which suggest genetic complexity (Sherman et al. 1985). Biochemical and cell biological studies have failed to provide an assay for genetic heterogeneity in MBS and linkage analysis is the only available method. Such analysis, however, is complicated by the incomplete penetrance of the disease in males and the variable penetrance and expression of the defect in heterozygous females. We have used a new approach to test the heterogeneity of recombination between MBS and the coagulation factor IX gene or the anonymous probe 52A in a group of nine families who have sought genetic counselling at Guy's Hospital. We find that both our families alone and our families plus apparently complete samples of pedigrees reported in the literature, separate into two groups: one tightly and one loosely linked to factor IX. In the combined family sample these represent respectively 0.3 and 0.7 of the total and show recombination fractions of 0.0-0.15 and 0.25-0.5. Furthermore, the families with non-penetrant carrier males show tighter linkage to factor IX than the others, thus confirming the suggestion of a systematic difference among MBS families in the recombination between the disease and the factor IX locus. By contrast, no significant differences were found in the recombination between 52A and factor IX in the two groups of MBS families or in these families versus those with Hunter syndrome examined in our laboratory. The causes of the linkage heterogeneity we describe are not known. At least two alternatives can be considered: The existence of two MBS loci or differences in the recombination between a single MBS locus and the factor IX gene. The association between incomplete penetrance and tight linkage to factor IX as well as the discontinuous variation in recombination fraction we have observed seem to favour the former alternative.