OBJECTIVE--To determine factors related to mortality within hospital after successful resuscitation from ventricular fibrillation outside hospital by a mobile coronary care unit manned by a physician. DESIGN--Retrospective review of records of patients resuscitated and admitted to hospital between 1 January 1966 and 31 December 1987. SETTING--Mobile coronary care unit, coronary care unit, and cardiology department. PATIENTS--281 patients (227 male), aged 14-82 (mean 58) successfully resuscitated from ventricular fibrillation outside hospital of whom 182 (65%) developed ventricular fibrillation before the arrival of the mobile coronary care unit. The aetiology of ventricular fibrillation was acute myocardial infarction in 194 patients (69%), ischaemic heart disease without infarction in 71 (25%), and other or unknown in 16 (6%). MAIN OUTCOME MEASURES--Death within hospital. RESULTS--There were 91 deaths in hospital (32%). Factors on univariate analysis significantly associated with increased mortality were patient age > or = 60 years, previous myocardial infarction or cerebrovascular disease, prior digoxin or diuretic treatment, collapse without prior chest pain or with pain lasting 30 minutes or less, defibrillation delayed by > or = 5 min, > or = four shocks required to correct ventricular fibrillation, left ventricular failure or pulmonary oedema and cardiogenic shock after successful defibrillation, and coma on admission to hospital. On multivariate analysis the most important factors (in rank order) were cardiogenic shock after defibrillation, coma on admission to hospital, age > or = 60 years and the requirement for four or more shocks to correct ventricular fibrillation. CONCLUSIONS--The in-hospital mortality of patients resuscitated from ventricular fibrillation outside hospital was related to patient characteristics before the cardiac arrest and to the immediate haemodynamic and neurological status after correction of ventricular fibrillation as well as to factors at the resuscitation itself. The in-hospital mortality of this study compares favourably with the results obtained by units staffed by paramedical workers and emergency medical technicians, although 35% (99/281) of the patients had ventricular fibrillation after the arrival of the mobile unit and defibrillation was thus rapid.