Studies of uniparental disomy and origin of nonmosaic trisomies indicate that both gain and loss of a chromosome can occur after fertilization. It is therefore of interest to determine both the relative frequency with which gain or loss can contribute to chromosomal mosaicism and whether these frequencies are influenced by selective factors. Thirty-two mosaic cases were examined with molecular markers, to try to determine which was the primary and which was the secondary cell line: 16 cases of disomy/trisomy mosaicism (5 trisomy 8, 2 trisomy 13, 1 trisomy 18, 4 trisomy 21, and 4 involving the X chromosome), 14 cases of 45,X/46,XX, and 2 cases of 45,X/47,XXX. Of the 14 cases of mosaic 45,X/46,XX, chromosome loss from a normal disomic fertilization predominated, supporting the hypothesis that 45,X might be compatible with survival only when the 45,X cell line arises relatively late in development. Most cases of disomy/trisomy mosaicism involving chromosomes 13, 18, 21, and X were also frequently associated with somatic loss of one (or more) chromosome, in these cases from a trisomic fertilization. By contrast, four of the five trisomy 8 cases were consistent with a somatic gain of a chromosome 8 during development from a normal zygote. It is possible that survival of trisomy 8 is also much more likely when the aneuploid cell line arises relatively late in development.