AbstractIn the economic evaluation of health care programmes, productivity costs are often estimated using patients' wages for the period of absence. However, the use of such methods for short periods of absence is controversial. A previous study found that short‐term absence is often compensated for during normal working hours and therefore does not lead to productivity losses. As such, the application of any approach almost certainly overestimates productivity costs. In this study, we examined the productivity costs for five different patient populations and one employee population, using the classical method and by identifying when extra effort was needed. In general, the results showed that productivity costs based on identifying extra effort were 25–30% of the classical estimates. For absences of just one day, productivity costs were relevant in only 17–19% of cases. For absences of two weeks or longer, productivity costs were relevant in 35–39% of cases. Measurement of the compensating mechanisms seemed to be valid, since there is considerable agreement between the opinion of supervisors and their employees about whether compensation covers productivity costs. There was much less agreement between supervisors and their employees on specific compensating mechanisms, however. The measurement of compensating mechanisms also seemed to be valid, because – as expected – different compensating mechanisms were reported for different occupations. In our study populations, compensating mechanisms appeared to differ with occupational characteristics, like part‐time work, managerial work and shift work. Copyright © 2004 John Wiley & Sons, Ltd.