This study aimed to validate that the computerized medical records within the General Practice Research Database (GPRD) can successfully identify congenital heart defects. Specifically, ventricular septal defect (VSD), Tetralogy of Fallot (TOF) and coarctation of the aorta (COA) were examined. Additional information from 1) medical record abstraction "questionnaires" sent to general practitioners and 2) maternal and infant free text comments fields within the GPRD was obtained. This information was compared to computerized medical records to understand which available components of the GPRD provide maximal information about congenital heart defects. This study also created case definitions of congenital heart defects within the GPRD and examined whether the prevalence rates of congenital heart defects for 2001-2003 obtained from the GPRD are similar to those obtained from two UK population-based systems, the National Congenital Anomaly System (NCAS) and the European Concerted Action of Congenital Anomalies and Twins (EUROCAT). Using practitioner questionnaires as the "gold standard," an overall positive predictive value (PPV) of the computerized medical record of 93% for the three defects was achieved. Of the 200 records, 44% contained infant free text information pertaining to the congenital heart defect of interest: the availability did not vary by specific defect. The proportion of infant defect information in the mother's free text was low. The prevalence of all heart defects from the GPRD was more than twice as high as in the NCAS and slightly higher than in the EUROCAT. All prevalence rate differences reached statistical significance (P=0.05). This study confirmed with a high positive predictive value that computerized GPRD medical records are sufficient to assess VSD, TOF and COA. It also demonstrated that the GPRD has the potential to produce a more complete estimate of prevalence of congenital heart defects compared to currently available UK national systems. The advantages of the GPRD as a medical records database including case ascertainment methods that do not rely on voluntary reporting, a theoretically unlimited duration of follow-up time in which a case can be detected and the availability of infant free text information make the GPRD a unique source for future studies of congenital heart defects.