OBJECTIVES: To use a stochastic all-cause, cause of death mortality model to determine the risk of cardiovascular (CVD) mortality in smokers compared with non-smokers in the UK allowing for competing causes of death. METHODS: The Sonata Vivo model adjusts population baseline mortality for known risk factors to calculate the mean age of death and most likely causes of death for an individual, and has been validated against long-term cohorts in the UK and USA. We used the model to calculate the likely outcomes for a 50 year-old male who smokes no, 10 or 20 cigarettes a day, with population average values for blood pressure, body-mass index, cholesterol and alcohol consumption. RESULTS: The mean ages of death were 82.7 years for the non-smoker, 78.7 years for a 10/day and 76.1 years for a 20/day smoker. At each year of life, smokers had a higher mortality from CVD than non-smokers. However, overall, there were fewer CVD deaths in smokers than non-smokers. This is an example of Simpson’s paradox, where an association between factors at a population level may be the opposite for each subgroup in the population. Smokers have increased mortality from cancer and respiratory disease, and on average die of these at a younger age than those who eventually die of CVD. Smokers also have fewer years of life than non-smokers in which CVD deaths can occur. CONCLUSIONS: Smoking increases the risk of many diseases, which have their greatest impact at different ages. Smokers have lower overall risks of CVD death as they are more likely than non-smokers to have died of cancer before the peak age for CVD deaths. The real-world effect of interventions for smoking-related diseases will be influenced by this impact of competing causes of death.