In 681 patients admitted to the coronary care unit (CCU) at Sahlgrenska hospital between 1 May 1983 and 31 May 1984, due to suspected acute myocardial infarction (MI), the hospital mortality and morbidity were related to whether the patients were transferred to hospital by ambulance or not. In the ambulance group acute MI developed in 48% (during the first 3 days in hospital) compared with 41% in the non-ambulance group (P = 0.10). The overall mortality rate was 10.4% in the ambulance group versus 3.8% in the non-ambulance group (P = 0.001). Corresponding figures for MI patients were 19.3% versus 9.1% (P = 0.02). In all, patients referred by ambulance had larger infarcts according to maximum serum enzyme activity and a higher incidence of congestive heart failure. Similar findings were observed when MI patients were analysed separately. On the other hand, the incidence of ventricular fibrillation, requirement for lidocaine, and the course of pain was fairly similar in the two groups. In a multivariate analysis, infarct size was the major independent predictor for early mortality rate. We conclude that patients who call for an ambulance due to suspected acute MI appear to have a different early mortality and morbidity pattern compared to those who do not. The most obvious observation was a higher early mortality. These patients therefore might be the most suitable candidates for early intervention studies.