OBJECTIVE Review of the results of surgical correction of atrioventricular septal defects (AVSD), identification of risk factors for mortality and failure of left AV valve repair and determination of the impact of cleft closure on postoperative AV valve function. METHODS Between 1975 and 1995, 121 consecutive patients (55 males, 66 females) underwent surgery for biventricular correction of AVSD with a median age of 1.2 years and a median weight of 7.6 kg. Sixty-five patients had a complete AVSD, 17 patients an intermediate type, and 39 patients a partial AVSD. The left AV valve (MV) cleft was closed in 53 patients (43.8%). The mean follow-up time is 7.2+/-4.6 years. RESULTS Actuarial survival of the whole group after 1 year was 80%, after 10 and 20 years 78 and 65%, respectively. There were 18 early deaths (7-day mortality, 10.7%; 30-day mortality, 14.9%) and eight late deaths. In a univariate analysis, risk factors for early or late death were diagnosis of complete AVSD (P=0.006), no cleft closure (P=0.024), postoperative complications (P<0.0001), age <1.2 years (P=0.017), weight <7.6 kg (P=0.002), PA/Ao pressure ratio >0.7 (P<0.0001), and ECC time >110 min (P=0.002). In the multivariate analysis, postoperative complications (P=0.003) and PA/Ao pressure ratio >0.7 (P=0.001) had parallel effects on the postoperative risk for mortality. Moderate or severe MV regurgitation was present in six patients (6.0%) in the first evaluation after discharge and in 20 patients (20.4%) in the most recent postoperative control. There were 25 reoperations in 17 patients, of which 15 had to be performed for MV regurgitation and two for MV stenosis. Freedom from reoperation was 91% at 1 year, 79% at 10 years, and 76% at 15 and 20 years. We could not identify a statistically significant risk factor for reoperation. CONCLUSIONS In patients with AVSD of various morphologies closure of the left AV valve cleft significantly improves outcome without affecting the need for reoperation. Risk factors for early and late death (multivariate analysis) were a pulmonary/aortic pressure ratio >0.7 and the occurrence of any complication after surgery. The concept of an early surgical AVSD correction before an increase in pulmonary vascular resistance and AV valve deformations occur would represent a better surgical option than a late correction as done in our series. Early correction allows for reduction of early mortality, superior long-term survival rates and a high freedom from subsequent valve degeneration.