The present study provides support for the utility of studying absence as a multi-dimensional criterion. Survey responses were collected from 194 bus drivers and paired with categorized archival absences. Seven absence indices were created and linked with three categories of predictors: (1) affective reactions to the work environment; (2) work-related perceptions; and (3) individual resource characteristis. The relationships between the multiple absence criteria and the three sets of predictors were examined both separately and combined using part canonical, and canonical correlation analyses. Affective responses fully mediated the influence of work perceptions on absence, and partially mediated the influence of individual resource variables. Redundancy coefficients and a rotated structure matrix were employed to identify two significant dimensions labeled, nonwork obligations and stress reactions, that linked the combined predictor sets with the set of absence measures. Together these dimensions accounted for 15 per cent of the variance in absence, with predictors differing in their explanatory power. Implications for the management of employee absence programs were discussed.