Serum unconjugated estriol concentrations before and after the oxytocin challenge test (stress test) were measured in 54 high-risk, obstetric patients. Six of the stress test results were positive, and eight were equivocal. Eleven patients demonstrated decrements in estriol of greater than or equal to 25% after stress tests, and the infants of six of these patients demonstrated perinatal morbidity. Three patients exhibited positive stress test results but had estriol concentrations which were unaltered by the stress test; all were delivered vaginally of healthy infants shortly after the stress test. The coefficient of correlation between the stress test and perinatal outcome was 0.58 (P less than 0.001). When the estriol and stress test results were combined to predict outcome, a multiple correlation coefficient of 0.71 was obtained (P less than 0.001). This increment represents a significant (P less than 0.001) and substantial (18%) improvement in the prediction of perinatal outcome. These findings indicate that the measurement of changes in estriol with the stress test significantly improves one's ability to define the fetus at risk.