We investigated to what extent psychosocial work characteristics predict sickness absence in a cohort of 890 human service professionals (84% women), followed-up for 3 years. We measured 16 different psychosocial work characteristics at baseline and analysed their associations with number of sickness absence days at follow-up using multivariate Poisson regression. In addition, we computed a psychosocial work environment index, summarizing eight psychosocial scales. Participants with exposure to violence and threats, high emotional demands, high demands for hiding emotions, low influence at work, low meaning of work, low quality of management, and role conflicts had an increased number of sickness absence days at follow-up, after adjustment for numerous confounders. Adjusting for sickness absence history caused little change in most effect estimates. Scoring in the most adverse quartile of the psychosocial work environment index was associated with a 71% increase in sickness absence days. Improving the psychosocial work environment index and eliminating exposure to violence and threats would have prevented 32% of all sickness absence days in the study population. In particular we found that exposure to violence was a strong predictor of absence. This study shows that a wide range of psychosocial work characteristics contribute to sickness absence in human service workers. Improving the psychosocial work environment might help to reduce sickness absence in this population.