Based upon unique Canadian administrative data from the years 1996 to 1999, this study examines the duration of absences from work due to injuries arising from workplace violence with a hazard model. We find that policing and nursing occupations, larger health care expenditures and more severe acts of violence are associated with longer absences from work. On the other hand, workers from larger firms have shorter absences from work. Our estimates are also quite sensitive to the inclusion of unobserved heterogeneity distribution, i.e., an individual specific random effect. This suggests that unobservable factors, such as stress and psychological or psychosomatic problems resulting from the workplace violence could have a large impact on the duration of work absences.