A population-based case-control study of non-Hodgkin lymphoma was conducted in the San Francisco Bay Area of California from 1988 to 1995. The study used Working Formulation histologic subtypes of non-Hodgkin lymphoma. Diffuse and immunoblastic large-cell lymphomas and all follicular lymphomas were combined to reflect two Revised European-American Lymphoma (REAL) classifications. Results were obtained from age- and sex-adjusted polytomous multivariable logistic models for 1,304 cases and 2,402 controls who were negative for human immunodeficiency virus. Statistically significant odds ratios for REAL large-cell lymphoma were decreased for receipt of five or more different vaccines (odds ratio (OR) = 0.65), use of calcium channel blockers (OR = 0.35), mononucleosis (OR = 0.55), plant allergies (OR = 0.72), and animal allergies (OR = 0.40). Odds ratios for REAL follicular lymphoma were decreased for use of nonsteroidal antiinflammatory drugs (OR = 0.67), hepatitis (OR = 0.36), and having had three or more bee or wasp stings (OR = 0.79) and were increased for heart disease (OR = 1.5). Similar elevated odds ratios for heart disease and Bacillus Calmette-Guérin vaccine and decreased odds ratios for nonsteroidal antiinflammatory drugs, plant allergies, and three or more bee or wasp stings were found for small lymphocytic lymphoma. These data provide direction for further investigation of immune function and genetic characteristics of subtypes of non-Hodgkin lymphoma in lymphomagenesis and warrant analyses in large pooled data sets.