Infant botulism results from the in vivo production of toxin by Clostridium botulinum after it has colonized the infant's gut. Epidemiologic and laboratory investigations of this recently recognized disease were undertaken to identify risk factors and routes by which C. botulinum spores might reach susceptible infants. Clostridium botulinum organisms, but no preformed toxin, were identified in six different honey specimens fed to three California patients with infant botulism, as well as from 10% (9/90) of honey specimens studied. By food exposure history, honey was significantly associated with type B infant botulism (P = 0.005). In California, 29.2% (12/41) of hospitalized patients had been fed honey prior to onset of constipation; worldwide, honey exposure occurred in 34.7% (28/75) of hospitalized cases. Of all food items tested, only honey contained C. botulinum organisms. On household vacuum cleaner dust specimens and five soil specimens (three from case homes, two from control homes) contained Clostridium botulinum. The known ubiquitous distribution of C. botulinum implies that exposure to its spores is universal and that host factors contribute importantly to the pathogenesis of infant botulism. However, honey is now an identified and avoidable source of C. botulinum spores, and it therefore should not be fed to infants.