Although a high prevalence of hepatitis C virus (HCV) infection among non-Hodgkin's lymphoma (NHL) patients had been reported, subsequent epidemiological studies conducted to examine a causal association between HCV and NHL have provided inconsistent results across studies. A strikingly positive association has been reported primarily from Italy and Japan, while no association was found in other regions of the world. To clarify the association between HCV and NHL, we conducted a systematic literature review. Eligible study designs were nested case-control studies, population-based case-control studies, and hospital-based case-control studies using non-cancer subjects as controls. The studies published through January 1991 to August 2003 were searched through Medline. Ultimately, 23 studies with 4049 NHL patients and 1,813,480 controls were identified. Summary statistics were crude odds ratios (ORs) comparing the anti-HCV seropositive and seronegative subjects. As we identified heterogeneity between studies, summary statistics were calculated based on a random-effect model. We did not find any evidence of publication bias. The major sources of variation were the use of blood donor controls and year of publication. The summary OR for NHL was 5.70 (95% confidence interval (CI), 4.09-7.96, P < 0.001). The subgroup analysis by phenotype showed a similar trend for B-cell (5.04, 95% CI: 3.59-7.06) and T-NHL (2.51, 95% CI: 1.39-4.56). In conclusion, we found a strongly positive association between anti-HCV seropositive test subjects and risk of NHL. Further biological studies examining this association are warranted.