In 1996, active surveillance in 5 Foodborne Diseases Active Surveillance Network (FoodNet) sites revealed up to a 9-fold difference in Escherichia coli O157:H7 (O157) infection incidence between sites. A matched case-control study of sporadic O157 cases was conducted in these sites from March 1996 through April 1997. Case subjects were patients with non-outbreak-related diarrheal illness who had O157 isolated from their stool samples. Control subjects were healthy persons matched by age and telephone number exchange. Overall, 196 case patients and 372 controls were enrolled. O157 infections were associated with farm exposure, cattle exposure, eating a pink hamburger (both at home and away from home), eating at a table-service restaurant, using immunosuppressive medication, and obtaining beef through a private slaughter arrangement. Variations in cattle exposures may explain a part of the regional variability of O157 infection incidence. O157 control measures should focus on reducing risks associated with eating undercooked hamburger, dining at table-service restaurants, and farm exposures.