The value of the contraction stress test (CST), although well documented in late pregnancy, has been questioned earlier in gestation. We have evaluated the reliability of the CST in 102 patients tested before 33 weeks' gestation. Eighteen patients with a positive CST had a significantly higher incidence of abnormal urinary estriol excretion (60% of patients), low Apgar scores (44%), growth retardation (39%), and perinatal mortality (277/1000) than did 84 women without a positive CST. Two stillbirths and 3 neonatal deaths occurred in the positive CST group. None of the neonatal losses was due to unnecessary premature intervention. Four patients with a positive CST showed no signs of fetal compromise, a false positive rate of 22%. Delivery was safely delayed an average of 6.1 weeks in patients without a positive CST. These findings demonstrate the predictive value of the CST and support its clinical application early in the third trimester.