Recent reports in selected patients have suggested that a reduced resting single-breath carbon monoxide diffusing capacity may be associated with a fall in arterial oxygen saturation during exercise. To determine if the diffusing capacity could serve as a screening test for changes in oxygen saturation in an unselected population, results of exercise studies were examined in 106 patients consecutively referred to an exercise laboratory. Nearly half of the patients underwent exercise testing to evaluate interstitial disease whereas the remainder were referred for unexplained dyspnea or for disability evaluations. Arterial desaturation was seen within all patient subgroups and was closely associated with reduced diffusing capacity. For detecting changes of 4 percent or more in oxygen saturation, a diffusing capacity of less than 50 percent of predicted gave the best combination of sensitivity (89 percent) and specificity (93 percent), whereas a diffusing capacity of 60 percent or less of predicted was 100 percent sensitive and 64 percent specific. For detecting lesser degrees of desaturation, sensitivities were slightly reduced but specificities were preserved. Thus, a diffusing capacity of less than 50 percent of predicted was associated with substantial arterial desaturation during exercise, whereas patients with a diffusing capacity of more than 60 percent of predicted had no desaturation during exercise. These results suggest that the resting diffusing capacity can serve as a screening test for exercise-induced hypoxemia in an unselected population.