Both fragile X females and mosaic males carry a methylated FMR1 allele in some cells and an unmethylated allele in other cells. Due to FMR1 protein produced in cells with an unmethylated allele, these individuals are expected to display a less severe cognitive phenotype than non-mosaic affected males. A larger range in cognitive abilities is expected for these individuals due to individual variation in X-inactivation or mosaicism ratios. These cognitive effects are well documented for females and have been suggested for mosaic males. Data on fragile X females suggest an approximately linear decrease in mean IQ score as a function of the fraction of cells with the mutation carried on the active X chromosome. Analysis of these data suggests that threshold effects in the relationship between IQ score and X-inactivation ratios are negligible, and that X-inactivation occurs randomly (with no preferential inactivation of either X chromosome) at a stage when the embryo consists of approximately 5 cells. A similar analysis of future data on mosaic males could yield estimates of the probability that a given cell will carry a methylated FMR1 allele and of the number of embryonic cells at the time that mosaicism is established. Distributions of IQ scores among a population of heterozygotes or mosaic male are predicted for several values of these parameters. These distributions include random contributions to IQ scores due to non-fragile-X-related effects, as well as X-inactivation or mosaicism ratios that vary from one individual to another.