The aim of the study was to examine the daily exposure times to hand-arm vibrations in Swedish car mechanics, to test a method for estimating the exposure time without observing the workers for whole days, and to use the results for predicting the prevalence of vibration-induced white fingers (VWF) by the ISO 5349-model. Six garages were surveyed. In each garage, 5-10 car mechanics were observed in random order every 30 seconds throughout working days. The daily exposure time for each mechanic was estimated from the fraction of the observations that the mechanic was exposed. A total of 51 mechanics were observed, most of them on two different working days, yielding estimates for 95 days. The median effective exposure time was 10 minutes per day (95% confidence interval 5-15 minutes, arithmetic mean 14 minutes, maximum 80 minutes), and most of the exposure time was attributable to fastening and loosening nuts. The within-worker and between-worker variability was high (total sigma2 0.99, geometric standard deviation of 2.7). Using the observed exposure time and data on vibration levels of the main tools in Swedish car mechanics (average weighted acceleration level of 3.5 m/s2), the model in ISO-standard 5349 would predict that only three percent of the car mechanics will suffer from VWF after 20 years of exposure. In contrast, a recent survey of VWF showed the prevalence to be 25 percent. The precision of the observation method was estimated and was found to be good for the group daily mean. On the individual level the precision was only acceptable if the daily exposure time was > or = 40 minutes. In conclusion, the daily exposure time was short and the vibration level was limited. Nevertheless, hand-arm vibrations cause VWF in a significant number of car mechanics. The method of observing workers intermittently seemed to work well.