Objectives: The purpose of this study was to investigate the differences in smoking rates according to the major occupational categories in South Korea. Methods: The study subjects were a weighted sample of 24,495 men and 26,121 women aged 25-64 from the 2003 Social Statistics Survey, which was conducted by the Korea National Statistical Office. Occupation was classified according to the Korean Standard Occupation Classification. We computed the age-standardized smoking rates according to gender and occupations after adjusting for the education level, marital status, and self-rated health. Results: For men, the smoking rate in elementary occupations was two times higher than that of clerks (OR=1.98, 95% CI=1.74-2.26). In general, a more prestigious job(professionals) correlated with lower smoking rates, and less prestigious jobs correlated with higher smoking rates, except for legislators, senior officials and managers. For women, smoking among service workers was 4.1 times higher than among clerical workers (OR=4.11, 95% CI=2.87-5.88). For women, their occupations, except elementary workers, and the unemployed, the retired and the armed forces, failed to show significant differences in smoking compared with the clerical workers. After adjusting for education, occupational differences in the smoking rate for men were attenuated in most occupations, except for legislators, professionals, and technicians. Further adjustment for marital status and self-rated health had a minimal effect on the occupational differences in the smoking rate for men. For women workers with service or elementary occupations, the ORs of smoking were attenuated with adjustment of the educational levels. However, the ORs of smoking were increased in workers with service, sales or elementary occupations, as well as for legislators, and the unemployed, the retired and the armed forces, after additionally adjusting for marital status. Conclusions: More prestigious jobs generally correlated with lower smoking rates in both sexes. The anti-tobacco policy should consider smoking rate differentials by occupations.