To investigate the pathophysiologic mechanisms of ventilator dependence, we took physiologic measurements in 28 patients with COPD and 11 postcardiac surgery (PCS) patients receiving long-term mechanical ventilation during a spontaneous breathing trial, and in 20 stable, spontaneously breathing patients matched for age and disease. After 40 +/- 14 min of spontaneous breathing, 20 of 28 patients with COPD and all 11 PCS patients were judged ventilator-dependent (VD). We found that in the 31 VD patients tidal volume was low (VT: 0.36 +/- 0.12 and 0.31 +/- 0.08 L for COPD and PCS, respectively), neuromuscular drive was high (P(0.1): 5.6 +/- 1. 6 and 3.9 +/- 1.9 cm H(2)O), inspiratory muscle strength was reduced (Pdi(max): 42 +/- 12 and 28 +/- 15 cm H(2)O), and lung mechanics were abnormal, particularly PEEPi (5.9 +/- 3.0 cm H(2)O) and lung resistance (22.2 +/- 9.2 cm H(2)O/L/s) in COPD. The load/capacity balance was altered (Pdi/Pdi(max) and Ppl/Ppl(max) > 0.4) and the effective inspiratory impedance was high (P(0.1)/VT/TI >/= 10 cm H(2)O/L/s). Failure to wean occurred in patients with f/VT > 105 breaths/min/L and 56% of patients with COPD with f/VT < 80 breaths/min/L. Those who failed despite a low f/VT ( < 80 breaths/min/L) either showed ineffective inspiratory efforts, which artificially lowered f/ VT (n = 8), or did not increase breathing frequency (n = 5), but P(0.1) and P(0.1)/VT/TI were as high as in other VD patients. In the 31 VD patients, Pa(CO(2)) increased during the weaning trial (+12.3 +/- 8.0 mm Hg). We conclude that in the presence of a high drive to breathe, the imbalance between increased work load and reduced inspiratory muscle strength causes respiratory distress and CO(2) retention. Noninvasive measurements (breathing pattern, P(0.1), P(0.1)/ VT/TI) may give better insight into weaning failure useful in clinical decision-making, particularly in patients with COPD not showing rapid shallow breathing (56% in this study).