We conducted a prospective study to evaluate whether lack of an adequate increase in diffusing capacity for carbon monoxide (DL(CO)) during exercise is associated with a greater postoperative complication rate after lung resection. We used the three-equation method (3EQ-DL(CO)), a modification of the single breath DL(CO) technique to determine DL(CO) during exercise in 57 patients undergoing lung resection at Vancouver General Hospital from October 1998 to May 1999. 3EQ-DL(CO) was determined during steady-state exercise at 35% and 70% of the maximal workload reached in a progressive exercise test. Maximal oxygen uptake (VO(2)max), DL(CO) at rest, and the increase in DL(CO) during exercise were compared in relation to postoperative complications. Patients with complications had lower resting values of DL(CO) (R-DL(CO)), a smaller increase in DL(CO) from rest to 70% of maximal workload expressed as a percent of the predicted DL(CO) at rest ([70% - R]-DL(CO)%), and a lower VO(2)max than did patients without complications. Results suggested that (70% - R)-DL(CO)% was the best preoperative predictor of postoperative complications; a cutoff limit of 10% was the best index to identify complications, yielding a complication rate of 100% in patients with (70% - R)-DL(CO)% < 10% as compared with a complication rate of 10% in patients with (70% - R)-DL(CO)% >/= 10% (sensitivity = 78%, specificity = 100%). Patients who do not increase their DL(CO) sufficiently during exercise ([70% - R]-DL(CO)% < 10%) have a greater complication rate after lung resection.