Multiple sources were used to identify maternal deaths and their causes in a study carried out in Jamaica. These sources of information included a review of all deaths of women aged 12 to 49 years and included those occurring in hospitals (on maternity, surgical and medical wards and in casualty departments); reported to coroners' offices and the police; on whom post-mortems were carried out at hospitals, public morgues and for the Ministry of National Security; obtained from interviews with public health staff in all parishes and which were registered with the Registrar General's Department. Some 193 maternal deaths were identified giving a maternal mortality rate of 10 per 10,000 live births. No one source independently identified all maternal deaths. Hospital in-patient records yielded 133 deaths (69%), death certificates 74 (38%). Deaths due to certain causes were far more likely to be identified from particular sources eg those due to clinical mismanagement (complications of anaesthesia and blood transfusion) from hospital in-patient records; while deaths from ruptured ectopic pregnancy were more likely to come from coroners', police and morgue records. It is concluded that using multiple sources to identify maternal deaths in developing countries is an effective method to identify all maternal deaths.