A prospective pilot study for early amniocentesis was conducted over 4 years including 279 early amniocenteses (EAs at 10–14 weeks' gestation) and 181 mid‐trimester amniocenteses (MAs at 15 weeks upwards). The study was performed with EA and MA utilizing the same proceduralists, techniques, and cytogenetics laboratory. Patients were offered either procedure and the cytogenetics laboratory was not prospectively informed of the gestational age of each sampled pregnancy. In the early amniocenteses, less fluid was sampled and there were trends for (i) less cells and clones being available for analysis, less successful cultures, and a slightly longer reporting time; (ii) more multiple insertions (12·9 per cent of specimens vs. 9·9 per cent) and more bloody fluids (5·4 per cent specimens vs. 4·4 per cent); and (iii) a higher rate of pregnancy loss (2·2 per cent of EA vs. 0·6 per cent of MA). The multiple insertion rates for both EA and MA were comparatively high and were related to an increased frequency of blood‐stained fluids. For EA specimens, the rates of amniotic fluid leakage, preterm delivery, and pregnancy loss were moderate and not significantly increased over the MA group. This study adds more weight to the view that EA is a reasonably safe procedure and a reliable alternative to chorionic villus biopsy to provide early prenatal cytogenetic diagnosis. © 1998 John Wiley & Sons, Ltd.