Amniotic fluid testosterone measured by radioimmunoassay (RIA) without chromatography (immunoreactive testosterone) seems not to be a definitive test for prenatal sex determination in all cases. In this study testosterone (T) levels measured by RIA with chromatography of the amniotic fluid samples were compared with immunoreactive testosterone (iT) values, to determine the predictive accuracy of the two methods. In 111 amniotic fluid samples between 15 and 19 weeks of gestation iT and T were measured parallelly. There are significant differences between iT- and T-means of both sexes (p less than 0.001). 95%-condifence limits of iT-values of the male and female fetuses are largely overlapping. In contrast, the overlap of 95%-confidence limits of the T-values is only minor. The measurement of testosterone with chromatography of the amniotic fluid samples shows for prenatal sex determination in over 90% accuracy. This result is due to the elimination of sex-specific differences in crossreacting steroids within the amniotic fluid of both sexes.