Indirect evidence, notably ecological comparisons and an association with skin cancer, links non-Hodgkin's lymphoma (NHL) with exposure to sunlight. We conducted a population-based, nationwide cohort study with exposure to outdoor work inferred from job titles reported in the population and housing censuses in 1960 and/or 1970 and by classifying each individual's work and home addresses according to latitude. Follow-up for cancer incidence was accomplished through record linkages with the virtually complete Swedish Cancer Registry. The cohort included all Swedish residents who were recorded as gainfully employed in both censuses. Altogether 4,171,175 individuals contributing 69,639,237 person-years accrued through 1989 were included in the analyses. We identified 10,381 cases of NHL, 4,018 cases of chronic lymphocytic leukemia (CLL), 11,398 cases of malignant melanoma (MM) and 11,913 cases of squamous cell skin cancer (SCC). We calculated age-adjusted relative risks for NHL, CLL, MM and SCC in strata based on estimated residential and occupational sunlight exposure. Interaction effects were considered for pesticide and solvent exposure. NHL, MM and SCC, but not CLL, were positively associated with increasingly southerly residential latitude, with stronger associations seen for skin cancer compared to NHL. Occupational sun exposure was not associated with the risk of developing any of the studied cancers. Pesticides and solvents also were not related to an increased risk of NHL, nor did these exposures enhance effects of residential or occupational sunlight exposure. Our results provide some support for an association of sunlight exposure with NHL incidence based on the associations seen using geographic latitude of residence as a proxy for exposure. Although type of occupation may be an imperfect index of the biologically relevant ultraviolet (UV) light dose, our data on individual exposure are not consistent with an important role of sunlight in the etiology of NHL.