This study was designed to compare the effectiveness of cervical assessment by either ultrasonography or bimanual pelvic examination in a program for preterm labor surveillance. Patients (n = 57) at risk for preterm birth were seen once a week for patient education, review of symptoms, and cervical evaluation. Cervical evaluation was assigned randomly to either ultrasonographic evaluation or pelvic examination. The groups did not differ in demographic or obstetric factors. The overall rate of prematurity was 18%. Preterm labor was more frequent with ultrasonographic evaluation (52%) than with pelvic examination (25%) (p less than 0.05). The group evaluated by ultrasonography received oral tocolytic agents (55%) more than the group that had pelvic examinations (21%) (p less than 0.02). The groups did not differ in infant birth weights, length of gestation, or neonatal mortality or morbidity. In this prospective randomized study of patients at risk for preterm birth, patients under surveillance by ultrasonographic assessment of the cervix did not fare better than those assigned to bimanual examination.