The total oestrogens detected by fluorescence in 24-h urine and the concentration of total oestriol detected in serum by radioimmunoassay were estimated in 51 women during the second half of pregnancy. The statistical analysis showed a linear regression between urinary excretion and serum concentration according to the equation y = -2.89 + 2.82x (y = serum oestriol). The correlation coefficient over all pairs of estimations (N =567) was r = 0.69, which is very highly significant (P less than 0.0001). The analysis of the individual pregnancies showed that the linear correlation coefficient was significant (P less than 0.05) in 35 cases, poorly significant (0.05 less than P less than 0.1) in 3 cases and not significant (P less than 0.1) in 5 cases. The correlation could not be calculated in 8 cases because the number of estimations was too small. These values were compared empirically (normal/pathological). There was agreement in 6 cases and no agreement in 2 cases. The results show on the one hand that the radioimmunological estimation of total oestriol in serum can be used in the same way as urinary excretion of oestrogens for monitoring the feto-placental unit. On the other hand, the study confirms that a "simple" fluorimetric Kober-Ittrich method can give at least the same clinical information as a radioimmunoassay of relatively high cost. It may be concluded indirectly from the results that a considerable diurnal rhythm or substantial day-to-day fluctuation in the production of oestriol are unlikely.