OBJECTIVE: The recommended criteria for public notification of a hepatitis A virus (HAV)-infected foodhandler include assessment of the foodhandler's hygiene and symptoms. In October 1994, a Kentucky health department received a report of a catering company foodhandler with hepatitis A. Patrons were not offered immune globulin because the foodhandler's hygiene was assessed to be good and he denied having diarrhea. During early November, 29 cases of hepatitis A were reported among people who had attended an event catered by this company. Two local health departments and the Centers for Disease Control and Prevention, in collaboration with two state health departments, undertook an investigation to determine the extent of the outbreak, to identify the foods and event characteristics associated with illness, and to investigate the apparent failure of the criteria for determining when immune globulin (IG) should be offered to exposed members of the public. METHODS: Cases were IgM anti-HAV-positive people with onset of symptoms during October or November who had eaten foods prepared by the catering company. To determine the outbreak's extent and factors associated with illness, the authors interviewed all case patients and the infected foodhandler and collected information on menus and other event characteristics. To investigate characteristics of events associated with transmission, the authors conducted a retrospective analysis comparing the risk of illness by selected event characteristics. To evaluate what foods were associated with illness, they conducted a retrospective cohort study of attendees of four events with high attack rates. RESULTS: A total of 91 cases were identified. At least one case was reported from 21 (51%) of the 41 catered events. The overall attack rate was 7% among the 1318 people who attended these events (range 0 to 75% per event). Attending an event at which there was no on-site sink (relative risk [RR] = 2.3, 95% confidence interval [CI] 1.4, 3.8) or no on-site kitchen (RR = 1.9, 95% Cl 1.1, 2.9) was associated with illness. For three events with high attack rates, eating at least one of several uncooked foods was associated with illness, with RRs ranging from 8 to undefined. CONCLUSION: A large hepatitis A outbreak resulted from an infected foodhandler with apparent good hygiene and no reported diarrhea who prepared many uncooked foods served at catered events. Assessing hygiene and symptoms s subjective, and may be difficult to accomplish. The effectiveness of the recommended criteria for determining when IG should be provided to exposed members of the public needs to be evaluated.