In this retrospective pilot study we examine the feasibility of establishing a confidential enquiry into why some patients die after emergency admission to hospital. After excluding those who died in the first hour or who were admitted for palliative care, pairs of physicians were able to collect quantitative and qualitative data on 200 consecutive deaths. Both physicians reported shortfalls of care in 14 patients and one of the pair in 25 patients whose deaths would not have been the expected outcome. In 25, the shortfalls of care may have contributed to their deaths. Major problems were delays in seeing doctors, inaccurate diagnoses, delays in investigations and initiation of treatment. They occurred mostly in those admitted at night. It is possible that establishing the correct diagnosis and starting appropriate treatment may have been delayed in 64% of the 200 patients. The headline figures appear worse than some previous external assessment studies but this study did concentrate on those in whom problems were more likely. Nevertheless, the frequency is too high to be overlooked. In this feasibility study we have demonstrated that it is practicable for local staff to collect and assess data in hospitals and that the types of problems identified are relevant to anyone planning how to organise emergency care. A larger definitive study should be performed.