INTRODUCTION: Noninvasive continuous positive airway pressure (CPAP) is recognized as an effective treatment for severe airway obstruction in young children. The aim of the present study was to compare a clinical setting with a physiological setting of noninvasive CPAP in infants with nocturnal alveolar hypoventilation due to severe upper airway obstruction (UAO) or bronchopulmonary dysplasia (BPD). METHODS: The breathing pattern and respiratory muscle output of all consecutive infants due to start CPAP in our noninvasive ventilation unit were retrospectively analysed. CPAP set on clinical noninvasive parameters (clinical CPAP) was compared to CPAP set on the normalization or the maximal reduction of the oesophageal pressure (Poes) and transdiaphragmatic pressure (Pdi) swings (physiological CPAP). Expiratory gastric pressure (Pgas) swing was measured. RESULTS: The data of 12 infants (mean age 10 ± 8 mo) with UAO (n = 7) or BPD (n = 5) were gathered. The mean clinical CPAP (8 ± 2 cmH₂O) was associated with a significant decrease in Poes and Pdi swings. Indeed, Poes swing decreased from 31 ± 15 cmH₂O during spontaneous breathing to 21 ± 10 cmH₂O during CPAP (P < 0.05). The mean physiological CPAP level was 2 ± 2 cmH2₂O higher than the mean clinical CPAP level and was associated with a significantly greater improvement in all indices of respiratory effort (Poes swing 11 ± 5 cm H₂O; P < 0.05 compared to clinical CPAP). Expiratory abdominal activity was present during the clinical CPAP and decreased during physiological CPAP. CONCLUSIONS: A physiological setting of noninvasive CPAP, based on the recording of Poes and Pgas, is superior to a clinical setting, based on clinical noninvasive parameters. Expiratory abdominal activity was present during spontaneous breathing and decreased in the physiological CPAP setting.