Purpose: To review available epidemiological data on the relationship of chronic obstructive pulmonary disease (COPD) to smoking. Methods: Epidemiological studies published up to 2006 relating smoking to the prevalence or incidence of COPD or closely related outcomes were reviewed. Meta-analytic methods were used to estimate the overall relative risk (RR). Results: A total of 117 independent studies, conducted mainly in Europe, North America, and Asia, were identified: 69 cross-sectional, 34 prospective, and 14 case-control studies. An association of COPD risk with smoking was consistently seen. Random-effects RR estimates compared to never smokers were 2.85 (95% confidence interval CI 2.60-3.13) for ever smoking and 3.48 (95% CI 3.07-3.94) for current smoking. Risk increased monotonically with increasing amount smoked with RRS of 2.77, 5.91, and 8.76 for approximately 5, 20, and 45 cigarettes per day and RRS of 1.25, 2.51, and 3.88 for approximately 5, 20, and 45 pack-years. There was marked heterogeneity between study-specific RR estimates. RRS were higher for studies of mortality and incidence, particularly for studies excluding asthma from their definition of COPD, and lower for studies of prevalent COPD, especially those based on lung function criteria. Conclusion: There is highly consistent evidence of an association of COPD with smoking, but the magnitude of the observed association varies.