Although maternal meiotic errors predominate in most studies of nonmosaic trisomy, studies of trisomy ascertained through confined placental mosaicism (CPM) have shown a high rate of somatic errors. However, origin of trisomy of many of the chromosomes involved in CPM has not been evaluated previously in cases ascertained through spontaneous abortions (SAs). Therefore, it was impossible to determine if the relative lack of meiotic errors in trisomy-CPM cases was a characteristic of the specific chromosome involved or due simply to ascertainment through a mosaic state. In the present study, parental and meiotic/somatic stages of origin of trisomy were determined in 89 SAs involving trisomy of chromosomes 2, 4 to 10, 12, 15, 17, and 20. Comparisons were then made to origin of trisomy in cases of confined and generalized trisomy mosaicism. Although somatic errors are generally more common in mosaic cases, this depends on the specific chromosome involved. The results suggest that there are chromosome-specific differences in the relative frequency of somatic chromosome gain or loss and/or the ability of an early somatic loss of one chromosome from a trisomic conceptus to "rescue" the pregnancy. As mean maternal age was less in the somatic than meiotic origin cases (P < 0.01), the age distribution of the study population should also influence the probability of detecting a somatic error. No phenotypic differences were apparent when cases were subdivided based on either parent or stage of origin of the trisomy.