IntroductionPatients with mullerian anomalies have higher rates of preterm delivery (PTD), but the true reason for PTD has not been elucidated. Transvaginal ultrasound (TVU) has proven an accurate test for the prediction of PTD, but has not been studied in this patient population.MethodsPatients with known mullerian anomalies were followed prospectively in pregnancy with TVU of the cervix between 14 and 24 weeks. The predictive value of TVU was evaluated using < 25 mm cervical length and/or > 25% funneling as definition for a short cervix. The primary outcome was PTD < 35 weeks.Results45 patients with mullerian anomalies were identified. 16 were excluded from analysis because of: prophylactic cerclage (6), twins (3), induced PTD (4), no follow‐up (1), still pregnant (2). Of the 29 pregnancies available for analysis, there were 8 with septate uterus (5 corrected), 7 with bicornuate uterus, 6 with didelphus uterus, and 5 with unicornuate uterus (3 were unknown type). 6 (21%) had a short cervix (3 of these received a therapeutic cerclage because of TVU short cervix), with 2 (33%) having PTD despite both receiving therapeutic cerclage. 23 (79%) did not have a short cervix, with 2 (9%) having PTD. Both of these patients had a didelphus uterus. The sensitivity, specificity, positive and negative predictive values for PTD were 50, 88, 40, 92%, respectively (RR 4.8, CI 0.9–26.5).DiscussionTVU seems to be predictive of PTD in patients with mullerian anomalies, with the possible exception of didelphus uterus. Therapeutic cerclage in patients with a mullerian anomalies with a short cervix on TVU may not prevent PTD.