Studies of employee tardiness, absence, and turnover generally adhere to one of five models: generalized withdrawal, which proposes positive intercorrelations among withdrawal behaviors; independent forms, which hypothesizes non-significant correlations among withdrawal behaviors; progression of withdrawal, which suggests that individuals engage in a hierarchically-ordered sequence of withdrawal; and the alternate forms and compensatory models which both imply substitutability of withdrawal behaviors. In an attempt to test the predictions of these theoretical models, the longitudinal lateness, absence, and turnover of 63 predominantly female (84 percnt) hospital employees was recorded bimonthly by the subjects' supervisors for a 48-week period. Subjects' job satisfaction, attitudes, health and perceptions of ease of mobility were obtained through self-reports and interviews. An analysis of the results, though failing to provide unequivocal support for any of the five models, did verify the interrelatedness of withdrawal behaviors. Single behavior models and general withdrawal models proved to be too simplistic for understanding withdrawal behavior. The progression of withdrawal model was most strongly supported, demonstrating a progression from lateness to absence to turnover. A 2-week lag between behaviors was observed. Little evidence of an ability to substitute behaviors was noted. Following the compensatory model, withdrawal behavior did have a feedback effect on health, but not on subsequent job satisfaction. (BL)