A case-control study was carried out on 97 consecutive preterm (< 37 weeks) infants to determine predictors associated with failure of nasal continuous positive airway pressure (CPAP) in the treatment of respiratory distress syndrome (RDS). Logistic regression analysis showed that only three risk factors were significantly associated with failed CPAP. These were: moderate or severe RDS (odds ratio: 5.9; 95 per cent confidence interval (CI): 2.2-16.0); septicemia during CPAP therapy (OR: 8.8; 95 per cent: CI 1.5-50.7); and pneumothorax during CPAP therapy (odds ratio: 6.9; 95 per cent: CI 1.1-41.7).