In a 12-month period there were 137 cardiac arrests in a district general hospital. Cardiopulmonary resuscitation was instituted within 3 minutes in 82%. Delay in 18% was due to equipment failure. Survival at 6 months was 12%. Of the 18% of inappropriate arrests, 42% had 'do-not-resuscitate' orders in the casenotes. Survival did not depend on age, sex, location, presence of an anaesthetist, experience of house officer, time of day or admission diagnosis. Survival was more likely in the presence of ventricular fibrillation and absence of intubation. Of the 32 arrest trolleys, 66% were geographically acceptable to the area they served and 9% had significant deficiencies (these were situated in patient waiting areas and were infrequently checked).