Objective: To assess the risk for coronary heart disease, myocardial infarction, cancer deaths, and all deaths associated with different smoking categories as determined by smoking status at a baseline examination only and at a baseline with reexamination 15-19 years later (persistent smokers). Material and methods: The participants were a random sample of 2930 men and 3084 women aged 34-61 years (when selected in 1967) invited for various standardized examinations under two periods, 1967-1972 and 1979-1991 and followed-up until the end of year 2001. The main outcome measures were clinical coronary heart disease, myocardial infarction, cancer deaths, and all deaths. Risk was calculated for each smoking category as determined by two assessments of smoking habits and also compared with the risk as determined by one baseline examination only. Results: Mean follow-up for men was 26 years (SD 9 years). For women the mean follow-up was 28 years (SD 7 years). There were substantial differences in hazard ratios (HR) and median lifetime in smoking groups as determined by one or two examinations. In men the greatest difference in hazard ratios was for cancer deaths (one examination: 2.80, two: 3.83) in women for total deaths (3.02 vs. 3.7). Loss of median lifetime was greatest in "heavy" cigarette smoking men (one examination: eight years; two examinations: 13 years), in women the corresponding figures were nine and 10 years, in "light" cigarette smokers, the figures for men were four and nine years, and for women four and six years. Conclusions: Middle-aged men smoking one or more packets of cigarettes per day shorten their life expectancy by 13 years and middle-aged women by 10 years. Only one baseline determination of smoking status with subsequent follow-up underestimates the health risk associated with smoking by 15-40% at least in populations where smoking prevalence is declining.