Audit is now an important component of the provision of acute cardiology services. In particular, the desire to administer thrombolytic treatment early in acute myocardial infarction has led to a reappraisal of admission procedures. Using records collected prospectively onto a computerized coronary care database for 36 months to December 1991, median delay before arrival at the emergency department and delay between admission and thrombolytic treatment was calculated. Of 1993 consecutive admissions to the coronary care unit, 816 patients had an initial diagnosis of myocardial infarction (later confirmed in 89.6 per cent), and 608 (74.5 per cent) of these received thrombolytic treatment. Overall median delay before arrival at hospital was 147 min. Randomization during the ISIS-3 trial significantly prolonged delays after arrival at hospital (64 vs. 50 min; p < 0.007). General practitioner referral delayed arrival at the emergency department (175 vs. 100 min self-referred; p < 0.0001) and was associated with similar hospital delay (54 vs. 55 min self-referred). Older patients ( > 65 years) presented later in the self-referred group (120 vs. 99 min for age < 65 years; p < 0.04), but there was no difference in the GP-referred group. Previous ischaemic heart disease did not predict type of referral. If delays before thrombolytic treatment are to be reduced significantly patients should be encouraged to seek early medical assistance by telephoning for an ambulance. Delays for patients arriving at the hospital following referral by a GP should be reduced by facilities for direct admission to the cardiologist.(ABSTRACT TRUNCATED AT 250 WORDS)