Although the effects of spirometric leaks on forced expiratory tracings have been noted in retrospect in epidemiologic studies, the present study was designed to examine the effects of leaks of known size on the tracings of subjects with varying degrees of airway obstruction. Such information is needed to assess the potential impact of spirometric leaks on epidemiologic studies using existing spirograms collected in the absence of spirometer leak tests. Spirometric leaks of 5 to 9, 22 to 30, 34 to 46, 69 to 97, and 150 to 190 ml/s were produced by drilling holes of five sizes into spirometer mouthpieces. Eight subjects were studied, of whom five had FEV1/FVC of 0.70 or greater ("normal"), and three had FEV1/FVC of less than 0.70 ("obstructed"). Although leaks of 34 to 46 ml/s or less were found to be invisible in the forced expiratory tracings, they caused statistically significant (p less than .05) FVC decrements in all eight subjects, with a mean decrement of -0.35 l in the normal subjects and -0.79 l in the obstructed subjects. In contrast, the 34 to 46 ml/s leaks caused no significant FEV1 decrements in seven of the eight subjects. Since marked FVC decrements occurred even with the smaller leaks, the FVC should not be analyzed in epidemiologic studies in the absence of frequent, documented leak tests. However, the present study indicates that the FEV1 may be robust enough for analysis if recorded in the presence of small spirometric leaks.