Lymphomas are one of the few cancers that have been increasing in incidence over the past decades. So far, only a few established risk factors have been identified, including immunosuppression and viral infections. Recent evidence suggests etiologic heterogeneity of different lymphoma subtypes. Smoking may affect risk differently, depending on the lymphoma entity. The European Prospective Investigation into Cancer and Nutrition was used to study the role of smoking in the etiology of lymphomas and individual subtypes within a prospective study. Information on baseline and lifetime tobacco smoking by 478,590 participants was collected between 1992 and 2000. Cox proportional hazards regression was used to calculate multivariate-adjusted hazard ratios and 95% confidence intervals. During 3,567,410 person-years of follow-up, 1,371 lymphoma cases (1,304 non-Hodgkin's lymphomas and 67 Hodgkin's lymphomas) were identified. Relative risk for smokers at recruitment was more than twofold higher for Hodgkin's lymphoma (hazard ratio = 2.14, 95% confidence interval: 1.18, 3.87) but was not elevated for non-Hodgkin's lymphoma (hazard ratio = 1.06, 95% confidence interval: 0.94, 1.19) and individual B-cell non-Hodgkin's lymphoma subtypes. In this prospective study, smoking appeared to increase Hodgkin's lymphoma risk consistently in both genders, whereas B-cell non-Hodgkin's lymphoma risk was not associated. Future analysis should involve viral biomarkers and genetic susceptibility markers to elucidate potential mechanisms of smoking-induced carcinogenesis, particularly for Hodgkin's lymphoma.