The perinatal mortality in 1498 patients with one or more previous cesarean section scars delivered between 1972 and 1982 was analyzed. Repeat elective cesarean section was performed in 654 (44%) patients and 844 (56%) were subjected to a "trial of scar." Successful vaginal delivery occurred in 702 (83%) patients and 142 (17%) had emergency repeat operations. There were 46 perinatal deaths, giving a perinatal mortality rate (PMR) of 30.3/1000. It was lowest in patients who electively had a cesarean section (10.6/1000). The corrected PMR was twice as high in the trial of scar group. The PMR for the overall hospital population (27,072 babies) during the study period was 22.5/1000. There were four perinatal deaths in association with dehiscence of the uterine scar. The PMR in trial of scar patients decreased from 40/1000 to less than 20/1000 without a major change in policy. Meanwhile the unit cesarean section rate increased from 5 to 10%, but the repeat section rate was consistent at around 38.5%. Regional analgesia was used in 192 patients, for repeat section in five and trial of scar in 187. Oxytocin was given to 546 (65%) patients. Scar rupture is considered the major contraindication to a trial of scar, but emphasis was not laid on the possible increased perinatal mortality with this procedure before the 1980s. In view of the improvement in the PMR and the added risk to the mother with cesarean delivery, we advocate a policy of trial of scar with informed consent in selected cases.