In the existing studies of rheumatic-fever recurrences, the criteria used for the diagnosis of carditis either were not defined or were frequently disparate from those used by other investigators.