Background Robust evidence is lacking regarding the clinical efficacy, safety and cardiopulmonary performance of perventricular closure. This study investigated the perioperative efficacy, safety and cardiorespiratory performance of perventricular closure of perimembranous ventricular septal defects (pmVSDs). Methods Operation-naive infants and young children aged 5-60 months with isolated pmVSDs were randomised to receive either standard open surgical or minimally invasive perventricular closure via direct entry into the ventricle with a catheter from a subxiphoid incision. The primary outcomes included complete closure at discharge, major and minor adverse events and the changes in perioperative cardiorespiratory performance from baseline. Complete closure was mainly analysed in the modified intention-to-treat (mITT) population, with sensitivity analyses for the ITT, per-protocol (PP) and as-treated (AT) populations (non-inferiority margin -5.0%). Results We recruited 200 patients with pmVSDs for this study (mean age 24.38 months, range 7-58 months, 104 girls), of whom 100 were randomly allocated to one of the study groups. The non-inferiority of perventricular to surgical closure regarding complete closure at discharge was not shown in the ITT (absolute difference -0.010 (95% CI -0.078 to 0.058)) and mITT populations (-0.010 (95% CI -0.069 to 0.048)), but was shown in the PP (0.010 (95% CI -0.043 to 0.062)) and AT populations (0.048 (95% CI -0.009 to 0.106)). Perventricular closure reduced the rate of compromising cardiac haemodynamics, electrophysiological responses, cardiomyocyte viability, respiratory mechanics, ventilatory and gas exchange function and oxygenation and tissue perfusion compared with surgical closure (all between-group P<0.05). Conclusions For infants and young children with pmVSD, perventricular closure reduced the rate of postoperative cardiorespiratory compromise compared with surgical closure, but the non-inferiority regarding complete closure should be interpreted in the context of the specific population.