A simple method for the determination of oestrone and oestradiol in human plasma is described and evaluated in terms of theoretical and practical errors. The oestrogens are separated on columns of Sephadex LH 20 and, after equilibration with antiserum, the unbound steroid is removed with dextran-coated charcoal. The mean total random theoretical error is calculated to be 22 %; the coefficients of variation of replicate analyses on pooled female plasma were from 13–18 % for both steroids; in male plasma the values were 17 % for oestradiol and 27 % for oestrone. The concentrations of oestradiol, expressed in pg/ml ± sd) in samples collected from healthy women during the menstrual cycle were 69 ± 56 (days 1–10); 126 ± 66 (days 11–20) and 99 ± 54 (days 21–30). Corresponding values for oestrone were 119 ± 46, 156 ± 41 and 156 ± 27.