To define risk factors for infant botulism, we performed a 2-year prospective case-control study of 68 laboratory-confirmed cases in infants living in the United States, outside of California. For each case patient, two control subjects were matched by date and hospital of birth or county birth records. By univariate analysis, breast-feeding (odds ratio = 2.9) and consumption of honey (odds ratio = 9.8) were associated with disease, but only 11 case patients (16%) had eaten honey. Decreased frequency of bowel movement (less than one per day for at least 2 months) was also associated with disease in infants 2 months of age and older (odds ratio = 5.2). Risk factors changed with the age of the patient at disease onset when analyzed by multivariate logistic regression methods. For infants less than 2 months old, living in a rural area or on a farm was the only significant risk factor (odds ratio = 6.4). For infants 2 months of age and older, breast-feeding (odds ratio = 3.8), less than one bowel movement per day for at least 2 months (odds ratio = 2.9), and ingestion of corn syrup (odds ratio = 5.2) were associated with disease. The severity of the disease was similar for breast- and bottle-fed infants. Clearly defined food exposures account for a minority of infant botulism cases. Preexisting host factors, such as intestinal flora and frequency of bowel movements, may be the most important risk factors for development of disease.