In this study, we analyzed the duration and determinants of prehospital delay in a group of inner-city patients hospitalized with suspected myocardial infarction. The average prehospital delay was 11.9 +/- 25.1 hours. Mean and median delays were similar for males (mean: 10.9 +/- 24.2 hours; median: 2.8 hours) and females (mean: 12.7 +/- 25.7 hours; median: 3.5 hours), but were longer for blacks (mean: 13.1 +/- 27.5 hours, P < .001; median: 3 hours, P = .06) and Hispanics (mean: 12.4 +/- 19.3 hours, P < .01; median: 4 hours, P = .07) than for whites (mean: 3.3 +/- 2.9 hours; median: 2 hours). Most of the observed delay was due to the time it took for patients to decide to seek medical care following onset of symptoms. Patients were more likely to arrive at the hospital within 4 hours if they thought their symptoms might be a heart attack (79% versus 41%, P < .01), if they believed that coronary heart disease was preventable (68% versus 42%, P < .01), and if they took an ambulance to the hospital (68% versus 47%, P < .01). The factor most strongly associated with early hospital arrival was the patient's belief that the symptoms might represent a heart attack; these patients were five times more likely to get to the hospital within 4 hours than others, independent of other factors. Interventions designed to decrease prehospital delay must focus not only on improving knowledge of symptoms, but also on identifying high-risk patients and increasing patient awareness of the benefits of early response and treatment.