The role of tobacco in the etiology of non-Hodgkin's lymphoma (NHL) was evaluated in a combined analysis of data from three population-based case-control studies conducted in four midwestern states of the United States: Nebraska, Iowa, Minnesota, and Kansas. Interviews were obtained from 1,177 cases (993 men, 184 women) and 3,625 controls (2,918 men, 707 women) or, if deceased, from their next-of-kin. Overall, there was no association between NHL and tobacco use (odds ratio [OR] = 1.0, 95 percent confidence interval [CI] = 0.8-1.1) or cigarette smoking (OR = 1.0, CI = 0.8-1.1). A slight negative association evident in analyses by intensity and duration of smoking was not present when interviews from proxy respondents were eliminated. There was a suggestion of a positive association between smoking and NHL among women (OR = 1.3, CI = 0.9-1.9), although there was no clear exposure-response relationship. This large case-control analysis provides no evidence that smoking is linked to the development of NHL among men. The possible role of smoking in the etiology of NHL among women needs further evaluation.