Objective : To estimate the prevalence of congenital heart disease from the 1990 student heart disease screening program. Methods : The heart disease screening program for elementary students was conducted in Kyonggi-do, in 1998. The subjects of the present study comprised the 40,402 students who attended the schools in the catchment area of a collaborative university hospital and who participated in the primary examination. The congenital heart disease (CHD) patients were initially identified through a questionnaire about prior medical history, and further through diagnostic tests & medical examinations in the secondary & the tertiary examinations. Certain assumptions were used in the estimation of the number of CHD cases among non-participants of the secondary & tertiary examinations. The overall prevalence of CHD was estimated by adding the CHD detection rates of the participants and the estimated prevalence of the non-participants. Results : Among the 40,402 primary participants, 1,655 were referred further, of whom 79.1% (1,309) participated in the secondary examination. Of these, 121 were referred to the tertiary examination, with a participation rate at this last stage of 80.2%. The positive predictive value (PPV) of the screening tools was the highest when the results of both EKG and the questionnaire were positive. Because 85.9% of the detected cases had a past history of CHD, PPV was higher when the selection criteria in the questionnaire included past CHD history than when it didnt. The CHD defection rate among the participants was 1.76 cases/1,000 and the presumed number of cases among the non-participants was 31; giving an estimated final CHD prevalence of 2.52 cases/1,000 (95% CI : 2.06-3.06). Among the identified cases of CHD, VSD (52.8%) was the most common, followed by PDA (9.7%), TOF (9.7%) & PS (9.7%). Conclusion : Because the characteristics of the non-participants differed from those of the participants, the estimation of prevalence was influenced by the participation rate. Of the detected cases, 85.9% had a past history of diagnosis or operation for CMD. These findings suggested that the prevalence estimated in this study may be an underestimation of the actual condition. Therefore, a birth cohort study is required in order to more accurately estimate the prevalence and the effects of the program.