Two hundred sixty-seven patients underwent surgical closure of a ventricular septal defect (VSD) over a 12-year period between January 1, 1970 and December 31, 1982. In 152 patients (57%) an associated lesion was present complicating the treatment of the primary lesion: 7 patients had multiple VSDs and among those with single defects, 189 were large (unrestrictive). The median age was 5.2 years ranging from 1 month to 46 years. Sixty-nine were infants under 10 kg of weight. The 30-day operative mortality was 8.6% (23 cases). The main cause of death was a low output syndrome in 16 cases. Thirty-two variables have been collected and their relationship with the death of the patients has been analyzed with univariate and multivariate methods. Logistic analysis has identified the independent significant incremental effect on hospital mortality of low weight (P greater than 0.00001), early operative date (P greater than 0.00001), multiplicity of defects (P = 0.0008), presence of major associated lesions (P = 0.019); the large size of the defect was only probably significant (P = 0.18). Inspection of the nomograms, relating the probability of operative death to weight and age in patients with a large defect, shows that the risk is significantly higher in the 3 kg baby (median age 3.3 months), particularly if major associated lesions are present. In infants with multiple VSDs the risk of operation remains, at the end of 1982, significantly higher than in single defects. However, our data suggest that the risk of open correction in patients over 12 kg of weight approaches that of single VSD if concomitant lesions do not complicate the surgical treatment.