The structure, process and outcome of cardiopulmonary resuscitation (CPR) at one hospital have been reviewed to determine where failings in the system could be improved, whether the existing training programme was adequate and well directed, and whether survival rates were reasonable. In the first seven months 143 reports were received; 49% of them were cardiac, 17% respiratory, and 24% mixed cardiorespiratory arrests. Overall, immediate survival was 43% of those with cardiorespiratory arrests. Seventy per cent of the immediate survivors were alive at 24 hours, and 69% of them survived to six weeks or beyond. The most important indicator for a successful outcome was the heart rhythm at the time of the resuscitation team's arrival, with 71% surviving from ventricular fibrillation, 20% surviving from apparent asystole, and only 9% surviving from electromechanical dissociation. No patient with the latter two rhythms survived to six weeks or left hospital. Neither the location or time of the arrest, nor the patient's age influenced the immediate survival. The audit revealed a need to improve telephone connections to the hospital switchboard, modify some equipment, and improve knowledge of the geography of the hospital site. Changes in emphasis in the training programme will alter the process of resuscitation. Outcome measures indicate that survival figures are comparable with published data, including recent data which act as a benchmark for quality control.