Purpose To determine which work arrangements, physical working conditions, and psychosocial working conditions are important risk factors for sickness absence. Methods Survey data on working conditions collected among the employees of the City of Helsinki during 2000 to 2002 (N = 6503, response rate 67%) were linked to the employer's sickness absence records for the subsequent 3 years. First occurrences of short-term (1−3 days), intermediate (4−14 days), and long-term (15 days or more) sickness absence episodes were examined by the use of proportional hazards models with Bayesian model averaging. Results Working overtime decreased the risk of short-term sickness absence by 19%. Heavy physical work load and hazardous exposures were consistently associated with increased sickness absence episodes of all lengths. The risk of intermediate and long-term absence episodes was increased by 24% to 28% per one standard deviation increase in physical work load. Low job control in women and job dissatisfaction in men increased the risk of sickness absence episodes of all lengths. Conclusions Heavy physical work load and hazardous exposures had the strongest associations with sickness absence. Furthermore, low job control in women and job dissatisfaction in men were consistently associated with increased risk of sickness absence. Systematic differences in risk factors for absence episodes of different lengths were not found.