OBJECTIVE: To compare knowledge and reported incidence of cardiac-related symptoms among patients with nurses' knowledge and estimated incidence of symptoms in this patient group. DESIGN: Retrospective collection of patient data by means of postal questionnaire and postal survey of hospital nurses. SETTING: Five non-randomly selected hospitals in northeast England. SUBJECTS: One hundred seventy-five male patients with a first uncomplicated myocardial infarction (MI), cared for by 168 registered nurses working in coronary care or medical wards. OUTCOME MEASURES: Reported and estimated occurrence of 11 common cardiac symptoms in the 3 weeks after MI and 15-item knowledge and 10-item misconception scales about cause of and recovery from MI. RESULTS: The percentage of nurses who correctly estimated the incidence of symptoms was low 25% of nurses did not make any correct estimates, and the mean number of correct estimates was not associated with nurses' experience or qualifications. The overall mean score for the knowledge scale was 9.6 (SD 1.9) for nurses, which was significantly higher (t = 7.5, p < 0.001) than that for patients (mean 7.9, SD 2.3); the nurses' score was not significantly associated with experience or place of work. For the misconception scale the mean score for nurses was 6.8 (SD 1.5), which was significantly higher (r = 6.85, p < 0.001) than the mean score for patients (5.5; SD 2.6); nurses working in specialized cardiac wards and more experienced nurses had significantly higher scores. CONCLUSIONS: Nurses should be provided with sound knowledge on cardiac symptoms and risks so that they can educate patients accordingly and, in particular, can correct misconceptions about the condition, prognosis, and appropriate lifestyle changes. Advanced-training personnel should recognize the need to enhance nurses' skills in patient education and rehabilitation; the importance of these skills also should be recognized in first-level training.