Historical data show that in Victoria birth defects have accounted for approximately 25% of all perinatal deaths. Terminations of pregnancies (TOPs) for birth defects occurring at > or =20 weeks gestation are included in the population-based perinatal data collection. These are classified as stillbirths or neonatal deaths. Some would have survived the perinatal period if no termination had taken place, and as a result they have the effect of increasing the perinatal mortality rate (PMR). Conversely, TOPs <20 weeks gestation, of fetuses with lethal birth defects that would have resulted in a perinatal death, are not included in the statistics and therefore reduce the PMR. The aim of this study was to examine the effect on the PMR of TOPs following the prenatal detection of birth defects, taking into account the severity or 'lethality' of the birth defects. Data on live births, stillbirths, neonatal deaths and TOPs carried out because of a birth defect were collected from the Victorian Birth Defects Register (BDR) for 1989-2000. Birth defects were categorised into three groups, according to the estimated likelihood of a baby with that condition dying in the perinatal period: a 'lethal' birth defect was one where there was >50% likelihood of death, 'possibly lethal' 15-50% and 'non-lethal' less than 15%. Based on these 'lethality' groups and associated assumptions about average survival rates beyond the neonatal period, the PMR was recalculated. TOPs for 'non-lethal' birth defects at > or =20 weeks gestation increased the PMR by 3.8%. TOPs for 'lethal' birth defects <20 weeks decreased the PMR by 14.4%. The net effect on the overall PMR from TOPs for birth defects was a 10.6% decrease.