A few epidemiologic studies have suggested that blood transfusion may be a risk factor for non-Hodgkin's lymphoma. The authors tested this hypothesis in a population-based, case-control study, using pathologically verified non-Hodgkin's lymphoma cases and transfusion documented via medical records. In 221 age- and sex-matched case-control pairs from Olmsted County, Minnesota, in 1975-1993, the authors observed an odds ratio of 0.84 (95% confidence interval 0.50-1.41) for history of transfusion and non-Hodgkin's lymphoma. There also was no apparent association between transfusion and non-Hodgkin's lymphoma in any subgroup analysis. Results do not support the hypothesis that blood transfusion contributes to the occurrence of non-Hodgkin's lymphoma.