Based on comprehensive administrative register data from Norway, we examine the determinants of sickness absence behavior; in terms of employee characteristics workplace characteristics, panel doctor characteristics, and economic conditions. The analysis is based on a novel concept of a worker's steady state sickness absence propensity, computed from a multivariate hazard rate model designed to predict the incidence and the duration of sickness absence for all workers. Key conclusions are i) that most of the cross-sectional variation in absenteeism is caused by genuine employee heterogeneity; ii) that the identity of a person's panel doctor has a significant impact on absence propensity; iii) that sickness absence insurance is frequently certified for reasons other than sickness; and iv) that the recovery rate rises enormously just prior to the exhaustion of sickness insurance benefits.