Data on smoking and mortality from the Swedish Twin Registry were analysed as a prospective cohort study and as a co-twin control study. The twin method involves control of genetic and early environmental factors and thereby a general control of the nested factors that may act as confounders, adjustments not obtainable in ordinary study designs. In the cohort analyses the following relative risks for cigarette smokers were found for men and women, respectively: death all causes 1.4 (90% Cl 1.3; 1.5), 1.4 (1.3; 1.5), CHD death 1.4 (1.3; 1.7), 1.6 (1.3; 2.0), lung cancer 19.7 (9.1; 42.7), 5.1 (3.0; 8.7), and other cancers 1.2 (1.0; 1.4), 1.2 (1.0-1.4). The comparison of deaths in cigarette-smoking twins and their non-smoking co-twins gave the following risk estimates for monozygotic (MZ) men: death all causes 1.6 (35 versus 22 first deaths), CHD death 2.8 (11 versus 4). The results for dizygotic (DZ) males and for females were in agreement. Four lung-cancer deaths occurred in MZ and 17 in DZ smoker twins while the non-smoker co-twins showed two such cases (DZ women). Other cancer deaths did not occur more often in the smoker than in the non-smoker twin. The impact of smoking on mortality, CHD death and lung cancer is also valid among smoking discordant twins.