During the course of a population-based epidemiologic study of congenital syphilis, we found discrepancies and problems of validity in the case definitions of congenital syphilis of the Centers for Disease Control and Prevention and other experts. We analyzed these problems and determined their impact on case classification in our study. The most important problem that we found was the case definitions' lack of a valid scientific basis for the classification of symptom-free infants born to mothers who have been treated for syphilis but have uncertain infection status (286 infants in our study). The classification of these infants is based on diagnostic tests whose sensitivity and specificity are unknown. In our study, we found that results of some tests were rarely positive and that values for others were similar to those in uninfected infants. We believe that symptom-free infants of treated mothers of uncertain infection status should be classified as cases, pending the development of better diagnostic tests for congenital syphilis. The economic impact of treating these infants can be lessened by obtaining fewer diagnostic tests and by use of one injection of penicillin rather than a 10- to 14-day course, an approach suggested by a literature review. Use of infants' diagnostic tests to classify any other group of symptom-free infants does not seem appropriate, even though this is often recommended.