The relationship between 6 chronic respiratory symptoms and the performance of an excessively variable FEV1 (test failure) was examined among 8,522 white adults in 6 U.S. cities. A total of 747 (8.9%) performed an excessively variable FEV1 according to the American Thoracic Society criterion. After adjusting for smoking, age, and city of residence in 6 separate logistic regression models, the odds ratios for FEV1 failure among men were 2.32, 1.39, 1.40, 1.82, 2.61, 1.92 for moderate breathlessness, chronic cough, phlegm, wheeze, asthma, and recurrent chest illness, respectively. Among women, FEV1 failure was significantly associated with moderate breathlessness, chronic phlegm, wheeze, and asthma with odds ratios of 1.55, 1.45, 1.62, and 1.95, respectively. When all symptoms were evaluated simultaneously in a single logistic regression model, only breathlessness and asthma remained associated with FEV1 failure; odds ratio = 1.97 for asthma and 2.03 for breathlessness among men and 1.53 for both asthma and breathlessness among women. The 11-yr mortality experience of subjects with test failure, as defined by 2 different criteria, was compared to that of the quartile of the cohort with the highest cross-sectional test results. After adjusting for age, gender, and smoking, the relative risks of mortality were 1.62 and 1.98 for subjects with an FEV1 failure as defined by the ATS and 6-Cities criteria, respectively, and 1.99 and 1.90 for the groups with FVC failure as defined by the 2 criteria. Thus test failure is almost as strong a predictor of mortality as poor FEV1.