Scoliosis is associated with progressive restrictive lung disease and an increased risk of pulmonary complications following surgical correction. Identification of higher risks for prolonged postoperative mechanical ventilation (MV) improves postoperative care. Our objective was to determine if preoperative pulmonary function tests (PFT) predict prolonged postoperative MV (defined as MV >or=3 days). We correlated preoperative PFT (forced expired volume in 1 sec, FEV1; vital capacity, VC; inspiratory capacity, IC; maximal inspiratory pressure, MIP; total lung capacity, TLC; and residual volume, RV) and postoperative MV days in 125 patients who had scoliosis surgery (aged 13.7 +/- 3.0 (SD) years) from January 1990-July 2001. We had 71 male and 54 female patients. Scoliosis types were 13 congenital, 27 idiopathic, 57 neuromuscular, 23 syndrome/tumor, and 5 kyphoscoliosis. Forty patients (32%) had postoperative MV >or=3 days. Independent factors likely requiring postoperative MV >or=3 days were neuromuscular scoliosis (P < 0.001) and FEV1 <40% predicted. Independent factors most likely were: neuromuscular scoliosis with preoperative FEV1 <40% predicted (P < 0.01). Independent factors most unlikely were: idiopathic scoliosis (P < 0.002). VC <60% predicted, IC <30 ml/kg, TLC <60% predicted, and MIP <60 cm H2O correlated with postoperative MV >or=3 days (P < 0.05). We found no association between RV and postoperative MV. FEV1 <40% predicted, VC <60% predicted, IC <30 ml/kg, TLC <60% predicted, MIP <60 cm H2O, and neuromuscular disease each correlated with prolonged postoperative MV. Neuromuscular disease or a preoperative FEV(1) <40% predicted were more likely, and older children with neuromuscular disease and FEV1 <40% predicted were most likely to require prolonged postoperative MV (P < 0.01). Clearly FEV1, and possibly VC, IC, TLC, and MIP, may increase accuracy in predicting the need for prolonged postoperative MV.