Clinical and morphometric features such as ventricular septal defect (VSD) size and location may determine outcome in infants with an isolated VSD. However, no currently available data allow quantitative estimation of the probability of spontaneous closure or surgery in individual patients. To identify independent predictors of outcome and to quantitate the probability of spontaneous closure and surgery in patients with isolated VSD, we studied 156 consecutive infants who had a diagnosis of an isolated VSD between January 1, 1988, and December 31, 1990, and who were subsequently monitored for 28.5 +/- 15 months. Of the 149 patients with membranous (n = 100) and muscular (n = 49) defects who were studied, 46 (31%) patients had spontaneous closure, and an additional 37 (25%) patients underwent surgical repair. Univariate analysis identified defect cross-sectional area indexed to body surface area, location in the muscular septum, presence of Down syndrome, and in membranous defects the presence of aneurysmal tissue as potential predictors of spontaneous closure or surgery. Multiple logistic regression analysis with these candidate variables identified indexed defect cross-sectional area as an independent predictor of spontaneous closure and surgery (p < 0.001). An inverse nonlinear relationship was seen between indexed VSD area and the probability of spontaneous closure (probability = (1 + e[-1.74 + 4.57CSA])-1 and a positive nonlinear relationship between indexed VSD area and the probability of surgery (probability = (1 + e[3.39 - 2.31CSA])-1). Muscular defects were more likely to close spontaneously than membranous defects (odds ratio 2.6, 95% CL = 1.01 - 6.8, p = 0.04).(ABSTRACT TRUNCATED AT 250 WORDS)