OBJECTIVES: To measure the impact on survival of being exposed to asbestos cement dust. METHODS: Survival of 866 asbestos cement workers and 755 controls was studied with Cox's proportional hazards regression models with age as the basic time variable. The effect of cumulative exposure up to the age of 40 was investigated in an internal analysis of 635 asbestos cement workers who had dose estimates. RESULTS: The death risk was higher for the asbestos cement workers than for the controls with a hazard ratio (HR) of 1.15 (95% confidence interval was 1.00 to 1.31). The increased risk found seemed to be confined to the period 20-40 years from start of employment. The estimates of the cohort effect were almost unaffected by adjustment for smoking habits. The estimates of the exposure effect rose with increasing dose (< 4 fibre-years/ml (f-y/ml): HR = 1.00, 4-9.9 f-y/ml: HR = 1.06, > or = 10 f-y/ml: HR = 1.35, for workers with at least five years of employment), and were higher when restricted only to deaths from malignant or non-malignant respiratory disease. However, none of the point estimates were significantly increased. Median age at death was two years lower in the high than in the low, exposure group. CONCLUSIONS: The results indicate that even a moderate asbestos exposure may shorten the median duration of life in an exposed population. Compared with the estimated effect on duration of life from ever being a smoker, that of ever being an asbestos cement worker was less, although that of having a high exposure was similar.