Fecal and serum specimens taken from 30 cases of sudden infant death and from eight cases of nonsudden infant death that were diagnosed at a single facility in King County, Wash., were examined for the presence of Clostridium botulinum organisms and toxin. Organisms, but not toxin, were recovered from a fecal specimen in one case of sudden infant death, results that parallel those from studies previously reported by investigators in California. Studies made in our laboratory of a nonfatal case of infant botulism revealed that an estimated 366,000 mouse minimal lethal doses of toxin were excreted in feces collected by purging the infant. Organisms and toxin were excreted for at least 15 days after the infant was hospitalized. Observations made in our laboratory of atypical responses in mice to both fecal and serum extracts, coupled with recently described experiments in which mice were used as an animal model for infant botulism in humans, provide a biologically plausible foundation for the hypothesis that C. botulinum may be implicated etiologically in some sudden infant deaths. Additional microbiologic, physiologic, and toxicologic data are needed to adequately test this hypothesis.