Longitudinal changes in IQ among mentally retarded (MR) fragile X [fra(x)] males have been reported previously. While age is associated with decline in IQ, not all males are so affected. This suggests that there may be more than one subtype of affected fra(X) male. Therefore, we examined the distribution of standardized difference scores (Zdiff) in IQ to determine if subjects were from an admixture of at least 2 populations. Cluster analysis of Zdiff scores was used to partition subjects into 2 groups. Goodness-of-fit tests indicated that scores were more likely to come from an admixture. Discriminant functions (DF) were calculated to determine predictive validity of Zdiff scores. To eliminate the effect of skewing, a power transform was applied to Zdiff scores and DFs recomputed. Zdiff and transformed scores provided similar results. The mean and variance for one group showed no differences in test-retest scores as would be expected from examining any population while the mean for the second group indicated significant decline in IQ nearly 4 standard errors below the first test score. These results suggest that there may be clinical evidence for 2 types of fra(X) mutation: One which causes MR but is static, and a second mutation which causes MR but is dynamic and contributes to an apparent longitudinal decline in cognitive function.