Granite workers lost to follow-up during a 5-yr study of pulmonary function were examined. We found that in workers who had left the industry for other employment, forced expiratory volume in one second (FEV1) values declined at a faster rate than in those who remained employed or retired during the study period. We investigated the application of a currently recommended repeatability criterion for pulmonary function tests by looking at different patterns of failing to satisfy the criterion (PFT-failure) over multiple testing sessions. Workers with persistent PFT-failure were found to have faster rates of FEV1 decline than did subjects with none or only intermittent failures. In a regression analysis, job termination was found to be more strongly associated with a steep FEV1 slope when it had been preceded by a pattern of persistent PFT-failure.