1. A study of the precision of clinical estriol- and total estrogen determinations in late pregnancy urine was carried out in collaboration with 26 clinical laboratories in the Netherlands and one laboratory in Suriname. 2. Ten urine samples were circulated twice with an interval of 2 weeks. 3. It was shown, that repeated analysis of the same sample in the same clinical laboratory in different assays can yield differences in results up to 40%. This interassay variation can be regarded as the main source of uncertainty of results of clinical estriol and total estrogen determinations. 4. The quantitative differences between results of total estrogen methods and a gas chromatographic method, which measures only estriol, were shown to be caused primarily by the lower recovery of the glucuronide of estriol (both native and added) in the latter method. 5. As expected, methods based on the principle of Ittrich (1960), Acta Endocrinol. 35, 34-48) proved to be more susceptible to the disturbing influence of glucose than a gas chromatographic method that measured estriol specifically. 6. Various recommendations to improve the precision of clinical estrogen determinations in pregnancy urine resulted from this study.