Background The aim was to investigate whether perceived work stress, overcommiment, balance in everyday life, individual factors and self-rated health in combination predict work ability among women and men in the public sector in Sweden. Methods A sample was randomly selected from the employee records of the participating public health care organisation in Western Sweden. In total, 2223 employees were included and answered a postal survey twice, at a 2 year interval. The survey included

more »

... ns about work ability, perceived work stress, overcommitment, balance in everyday life, individual factors and self-rated health. Odds ratios with 95% confidence intervals for work ability were estimated using logistic regression. Results Imbalance in everyday life and overcommitment predicted reduced work ability in women and imbalance in everyday life and low educational level predicted reduced work ability in men. However, when poor self-rated health was added to the models this was the strongest predictor of work ability for both genders. Conclusion A combination of poor self-rated health, imbalance in everyday life, and overcommitment predicted reduced work ability. This multifactorial nature of work ability should be taken into account in health promotion programmes.