'Under-reporting' of deaths to the coroner has significant implications for the identification and investigation of preventable deaths. In extreme cases, it may even be a symptom of the system failures that allowed cases such as Harold Shipman, Australia's King Edward Memorial Hospital, the alleged incidents at the Bundaberg Hospital and the Bristol Royal Infirmary to persist. Several initiatives in Australia and the UK are currently reviewing the coroner's system in light of the recommendations made by the Luce report and the Bundaberg Hospital inquiry to consider whether the coroner's system effectively meets the needs of our society, including the healthcare sector. Reporting of deaths to the coroner is a key issue for consideration in this debate.This study's primary aim is to identify the number of deaths in the hospital setting that meet the reporting criteria set out by the coroner's Act, Victoria 1985 ('reportable deaths').This study utilized a method of retrospective structured medical record review of in-patients who died between 1 January 2002 and 30 June 2003 at two major public hospitals in Victoria, Australia.In total, 229 cases (95.4% of records requested) were included in this review (120 from Hospital A and 109 from Hospital B). The number of cases at both hospitals meeting the coroner's reporting criterion was 58, of which, 22 (37.9%) were reported to the coroner.This study provides the first experimental evidence of significant 'under-reporting' of deaths to the coroner by hospitals. This is an important consideration for the reform initiatives currently underway. Better communication channels need to be fostered between doctors and coroners if coronial investigations are to be used effectively for reviewing deaths in hospitals.