The four-chamber view of the fetal heart has been advocated as an effective antenatal screen for severe forms of congenital heart disease (CHD). To evaluate the sensitivity of this test at a primary care level, a retrospective review was undertaken for the period 1988 to 1992 of ultrasonograms of fetuses referred for fetal echocardiography, mothers of neonates referred for cardiac assessment, and perinatal deaths attributed to CHD. The study centers serve a population base of 3.5 million persons. Sixty-nine cases of severe forms of CHD were reviewed. The diagnoses were chosen because all show abnormal findings on four-chamber view: 38 hypoplastic left heart syndrome, 13 pulmonary atresia with intact ventricular septum, and 17 variants of univentricular heart. The cases involved a total of 94 scans. These were reviewed by diagnosis and stratified by gestational age and year of scanning. No diagnosis was made at a gestational age younger than 16 weeks. From 16 weeks to term, overall sensitivity was 40% (32/80 scans); sensitivity improved with advanced gestational age (34% to 48% at 16 to 24 and > 24 weeks' gestation, respectively). There was no longitudinal evidence of a learning curve. Reasons for this low sensitivity are related to unfamiliarity with CHD, low incidence in small populations, and inadequate evaluation of the fetal heart. These problems can be resolved with continued training and direction from a major center.