Air leaks often result in alveolar hypoventilation in mechanically ventilated patients with neuromuscular disease. The primary objective of this study was to assess the feasibility, efficacy and tolerance of a ventilator equipped with an automated air-leak compensation system in a clinical situation. Fourteen neuromuscular patients with nocturnal air leaks during home ventilation were included in a prospective randomised crossover study. A modified VS Ultra ventilator was studied during two consecutive nights and patients were randomly ventilated with and without a leak-compensation system, respectively. Tolerance, minute ventilation, blood gas values, sleep parameters, and nocturnal oxygen saturation were assessed. Leak compensation significantly increased the mean inspiratory and expiratory tidal volumes (731+/-312 vs. 1094+/-432 ml [p=0.002] and 329+/-130 vs. 496+/-388 ml [p=0.006], respectively) and inspiratory and expiratory flows (51.7+/-8.2 vs. 61.8+/-12.4 l/min [p=0.016] and 63.3+/-26.2 vs. 83.3+/-37.8 l/min [p=0.013], respectively). The system acted by increasing both inspiratory time (from 1355+/-230 to 1527+/-159 ms, p=0.038) and inspiratory pressure (from 14.0+/-2.8 to 18.3+/-3.4 cm H(2)O, p=0.002). Leak compensation improved arterial PCO(2) (6.18+/-0.9 vs. 5.21+/-1.0 kPa, p=0.004), slow-wave-sleep latency (119+/-69 vs. 87+/-35 min, p=0.04), and tolerance. Air-leak compensation is feasible and may produce beneficial effects in neuromuscular patients. The automatic air-leak compensation system tested here should be evaluated in long-term efficacy and tolerance studies and compared to other ventilation modes capable of compensating for leaks, such as pressure support.