A survey of glycophorin A (gpa) in vivo somatic cell mutation in a population of 394 healthy people from 8 to 77 years of age (mean age +/- SD 41 +/- 15 years) revealed a subset of 37 individuals with stably elevated allele-loss and/or allele-loss with duplication variant erythrocyte frequencies (Vf) exceeding 30 x 10(-6). These 37 individuals with gpa outlier Vf are significantly older (P < 0.001) than the remainder of the larger study population from which they were drawn reflecting a highly significant increase in the prevalence of these individuals in the population beyond age 40 years. A study of hpt mutant frequencies (Mf) in the peripheral blood T-lymphocytes of 27 of these individuals, together with 15 matched control individuals with unremarkable gpa Vf, was undertaken to determine if these subjects also displayed elevated mutation frequencies at this independent locus indicative of globally elevated somatic mutation. The hprt Mf in these 27 subjects (geometric mean 11.5 x 10(-6)(dispersion interval 5.8 x 10(-6) to 22.8 x 10(-6)) was not significantly different from that observed in the 15 controls (geometric mean 12.1 x 10(-6)(dispersion interval 5.7 x 10(-6) to 25.5 x 10(-6)). These Mf are higher than typically reported values reflecting the older age distribution of these individuals (arithmetic mean age +/- SD 53 +/- 12 and 50 +/- 16 years for the subjects and controls, respectively). Taken together, these data suggest that several genetic mechanisms may be responsible for producing the gpa outlier Vf observed in these subjects. The observation that hprt Mf were not increased indicates that the majority did not arise by a genome-wide increased rate of somatic mutation detectable at both loci. The fixation and subsequent expansion of 'jackpot' mutations at the gpa locus occurring early in embryonic/fetal development also does not appear to be a predominant mechanism. Some cases may result from a stable over-representation of gpa variant cells, perhaps associated with a marked age-dependent decrease in the number of contributing erythroid stem cells in the bone marrow. The subset that displays elevated allele-loss with duplication Vf involving both gpa alleles may represent individuals with increased rates of somatic recombination. Elevations arising by this mechanism are not detected in the hprt assay, but could be confirmed using a autosomal locus in vivo somatic cell mutation endpoint such as the hla-a assay. Of primary biological significance, these results demonstrate that genetics/stochastic processes leading to the loss of heterozygosity of somatic cells occur ubiquitously in humans and in some individuals this level of somatic mosaicism can approach a frequency of 10(-3) at the gpa locus in erythroid lineage cells.