In-hospital management of out-of-hospital cardiac arrest is complicated by uncertainty about prognosis and the need to identify markers of adverse outcome in individuals surviving initial resuscitation. We sought to identify factors that predict in-hospital death among patients who initially survive out-of-hospital cardiac arrest. We investigated 346 consecutive cases of out-of-hospital cardiac arrest received by a single centre in Edinburgh, UK (270 cases examined retrospectively, 76 prospectively). Of the retrospective cohort, 246 cases were thought to be of cardiac origin. There were associations between in-hospital mortality and pre-arrest variables, resuscitation variables, and factors measured during admission. Crew-witnessed arrests were associated with low mortality; arrest rhythm (p < 0.001), resuscitation by a health professional (p < 0.05), conscious level on admission (p < 0.001), and requirement for ventilation (p < 0.05) independently predicted in-hospital mortality. A weighted prognostic scoring system based on three of these variables accurately predicted the likelihood of in-hospital death in the prospective test group. Further assessment of conscious level during admission with the Glasgow coma score predicted mortality rates in the study population, but coma did not predict a hopeless prognosis in individual cases unless it persisted for 72 h or more. Accurate prognostic assessment of out-of-hospital cardiac arrest survivors can be made from information available on admission. Of factors that independently predicted outcome, the skill of the resuscitator is most readily modified. This suggests that public training in resuscitation may reduce mortality rates.