This paper describes difficulties involved in conducting cross-sectional and longitudinal research on the effects of occupational stressors on the mental and physical health of veteran workers. The difficulties include (1) the disappearance, prior to the beginning of the investigation, of the casualties of job stress (Kasl, 1983) and (2) selection bias. The paper advances the view that an alternative design, one that (1) follows new workers longitudinally, (2) includes preemployment measures of mental and/or physical health, and (3) controls for nonoccupational stressors is warranted. The alternative design allows for (a) the assessment of the extent to which the findings are vulnerable to selection-based and event-proneness explanations (Dohrenwend and Dohrenwend, 1981) and (b) the estimation of relatively immediate effects of working conditions on health. The paper provides an example of such a design in a study of the effects of adverse working conditions on depressive symptoms in female teachers Ordinary least square regression and LISREL analyses suggest that adverse working conditions exert a relatively immediate effect on depressive symptoms in teachers, controlling for preemployment symptoms and other factors (age, race, social support, nonoccupational stressors, marital status, and social class or origin). Event-proneness and selection-based explanations were ruled out.