We examined the role of job satisfaction, emotional exhaustion, and job decision latitude as predictors of absenteeism among nurses at the aggregate (unit) level. The study was conducted in a university hospital. Usable data were collected from 625?respondents pertaining to 51?hospital units. We first used ecological logistic modelling of the form events/trials, where an event is an absence day and a trial is the sum of absence and worked days, to examine the contribution of our theorized predictors to absenteeism. Univariate analyses showed that absenteeism was significantly associated with emotional exhaustion (Wald chi-square = 190 . 8, p < .0001), job satisfaction (Wald chi-square = 377 . 5, p < .0001), and job decision latitude (Wald chi-square = 422,6, p < .0001). However, in multiple logistic regression, emotional exhaustion was not associated with absenteeism (Wald chi-square = 0.4,?n.s.), while job control (Wald chi-square = 231.0, p < .0001) and job satisfaction (Wald chi-square = 96.5, p < .0001) remained negatively associated with it. These relationships were further analyzed using a path-analytic approach through LISREL (version?8 . 72). Our specified theoretical model included both direct and indirect effects of predictor variables on absenteeism. This model yielded an excellent fit to the data, ?²?(1) = 0.068,?n.s., and revealed that job decision latitude had a significant effect on absenteeism (? = - .16, p < .001), but also influenced job satisfaction (? = .22, p < .001) and emotional exhaustion (? = - .40, p < .001). On the other hand, emotional exhaustion affected job satisfaction (? = - .61, p < .001) which in turn reduced absenteeism (? = - .15, p < .01). These findings highlight the key role played by job decision latitude and job satisfaction as determinants of absenteeism in nursing units. These findings are interpreted within the context of avoidance motivation as a key mechanism through which nurses self-regulate their sense of well-being and health in the workplace (cf.?Hackett &?Bycio, 1996) and as a step towards more severe forms of withdrawal behavior (cf.?Harrisson, Newman, &?Roth, 2006). The significance of these findings for future research on absenteeism in nursing units is also discussed.