A previous study (Stanhope et al. 1998) established that staff in two obstetric units reported less than a quarter of designated incidents to the units' risk managers. A questionnaire was administered to 42 obstetricians and 156 midwives at the same two obstetric units, exploring the reasons for low rates of reporting. Questions concerned their knowledge of their unit's incident reporting system; whether they would report a series of 10 designated adverse obstetric incidents to the risk manager; and their views on 12 potential reasons for not reporting incidents. Most staff knew about the incident-reporting system in their unit, but almost 30% did not know how to find a list of reportable incidents. Views on the necessity of reporting the 10 designated obstetric incidents varied considerably. For example, 96% of staff stated they would always report a maternal death, whereas less than 40% would report a baby's unexpected admission to the Special Care Baby Unit. Midwives said they were more likely to report incidents than doctors, and junior staff were more likely to report than senior staff. The main reasons for not reporting were fears that junior staff would be blamed, high workload and the belief (even though the incident was designated as reportable) that the circumstances or outcome of a particular case did not warrant a report. Junior doctors felt less supported by their colleagues than senior doctors. Current systems of incident reporting, while providing some valuable information, do not provide a reliable index of the rate of adverse incidents. Recommended measures to increase reliability include clearer definitions of incidents, simplified methods of reporting, designated staff to record incidents and education, feedback and reassurance to staff about the nature and purpose of such systems.