SummaryWe report data on diagnoses made on amniotic fluid specimens from 1977 to 1984 as reported to the New York State Chromosome Registry. The rate of allde novo(presumed mutant) abnormalities was about 2 per 1000 in about 61000 fetuses in which results are unlikely to be biased by the reason for amniocentesis (except for maternal age). This includes about 0·5 per 1000de novomarkers, about 0·5 per 1000 otherde novounbalanced, and about 1·0 per 1000de novobalanced rearrangements. In about 55000 fetuses in which rates of inherited abnormalities could be evaluated without apparent bias, the rate of all inherited rearrangement was about 2·9 per 1000. This includes about 0·3 per 1000 inherited markers, about 0·2 per 1000 other inherited unbalanced rearrangements, and about 2·4 per 1000 inherited balanced abnormalities. Only mutant markers showed a clear association with maternal age (37·6±2·7 in 24 casesv. 35·8±3·6 in controls). Inherited markers did not exhibit this trend (35·8±2·0 in 12 casesv. 36·4±2·8 in controls). Paternal age does not appear to account for the association. Among abnormalities of known origin, the ratio of mutant to inherited cases is for markers 64:36, for other unbalanced rearrangements 73:27, and for all balanced abnormalities 29:71. In a subgroup of about 55000 fetuses, of 263 total abnormalities there were 8 instances of apparent true somatic mosaics (5 mutant and 3 of unknown origin but almost certainly mutant). There were also 20 instances of markers in which presumptive somatic loss had resulted in mosaicism (10 mutant, 6 of unknown origin and 4 inherited) and 13 other instances of mosaicism associated with apparent somatic loss (9 mutant, 3 of unknown origin, and 1 inherited). The sex ratio (Y to non‐Y karyotypes) for all abnormalities detected was 228:210 (1·09), not different from controls. Only deletions (5:14) and ‘other’ unbalanced rearrangements (5:13) exhibited a suggestive deviation from this trend. The rates of mutant chromosome rearrangements reported from 1977 to 1983 showed no apparent time cluster, with the possible exception of a peak of markers in 1977, a trend that may be due to higher maternal age in this year. Among fetuses studied because of maternal exposure to putative mutagens there was a non‐significant excess of mutants (2·9–5·7 per 1000v. 1·7–2·2 per 1000) and a borderline significant excess of inherited rearrangements (8·6–11·5 per 1000v. 2·6–3·1 per 1000). The latter effect, if not due to chance, may be due to effects on segregation. Fetuses studied because of known or suspected fetal pathology were excluded from most of the analyses above. In those ascertained because of known or suspect fetal pathology, a group defined conservatively to diminish bias in the remainder, the rate of unbalanced abnormalities was notably higher than in the reference groups. For mutants the rate was 1·8–2·4 per 1000v0·8–1·1 per 1000 in controls and for inherited unbalanced abnormalities the rate was 1·2–1·8 per 1000v0·3–0·7 per 1000 in controls. For balanced abnormalities by contrast, the rates of mutants were 0·6–1·2 per 1000v0·9–1·1 per 1000 in controls, and of inherited abnormalities 3·6–4·2 per 1000v2·2–2·4 per 1000 in controls. Because of increasing use of serum alpha‐fetoprotein screening and early ultrasound screening of pregnancies, it is likely that in the future it will be increasingly difficult to derive unbiased rates of unbalanced abnormalities at amniocentesis, even after adjustment for maternal age. Balanced abnormalities may prove mor