OBJECTIVE: To determine whether there has been an improvement in ascertainment of birth defects cases ('case validity') by the Victorian Birth Defects Register (BDR) since an earlier study (conducted in 1993), to ascertain the accuracy of registered data ('item validity') and to investigate another possible source of notification. METHODS: The medical records were reviewed of 500 children born after 1 January 1993 who were consecutively admitted after 1 January 1999 to two paediatric teaching hospitals in Victoria. In addition, records of 200 children referred to a clinical genetics service were reviewed for children born after 1 January 1993 and who were seen in two periods: 2 months after 1 January 2001 and 2 months after 1 January 2002. The records from the hospitals and clinical genetics service were reviewed separately to determine whether children recorded as having a birth defect had previously been notified to the BDR. RESULTS: Twenty percent of the hospital records related to a child with a birth defect, as did 70% of the clinical genetics service records. Overall case validity for birth defect cases from the hospitals was 88%. There was 100% ascertainment for three of five categories. Sixty per cent of birth defects cases from the clinical genetics service had been notified to the BDR. When all diagnoses in matched cases were considered, item validity was 54%, however, if only primary diagnoses were included then 92% of cases had the same diagnosis. CONCLUSIONS: Overall case validity from the two paediatric teaching hospitals has significantly improved since our previous study. The addition of an extra data source from a clinical genetics service would identify new cases, particularly genetic disorders and developmental delay, as well as adding new diagnoses to existing ones. This study has highlighted the need to improve item validity, perhaps through routine education for all coders and notifiers on the inclusion and exclusion of specific associated conditions when notifying major birth defects.