Over a 34.5-month period, all admissions to 5 university hospital coronary care units were screened for eligibility for the Multicenter Investigation of the Limitation of Infarct Size (MILIS), an ongoing study of the effects of hyaluronidase, propranolol and placebo on myocardial infarct (MI) size. Of 3,697 patients with ≥ 30 minutes of discomfort that was thought to reflect myocardial ischemia who were assessed for the presence or absence of certain electrocardiographic abnormalities at the time of hospital admission, the electrocardiogram was considered predictive of acute Ml if ≥ 1 of the following abnormalities was present: new or presumably new Q waves (>- 30 ms wide and 0.20 mV deep) in at least 2 of the 3 diaphragmatic leads (II, III, aVF), or in at least 2 of the 6 precordial leads (V 1 to V 6 ), or in I and aVL; new or presumably new ST-segment elevation or depression of >- 0.10 mV in 1 of the same lead combinations; or complete left bundle branch block. In the screened population, the diagnostic sensitivity of the electrocardiographic criteria was 81 %, whereas the overall infarct rate in the total population screened was 49 %. The diagnostic specificity of these entry criteria was 69 % and the predictive value 72%. Further application of readily ascertainable study exclusion criteria (age > 75 years; qualifying symptoms commencing > 18 hours before presentation; previous MILIS participation; geographic, physical or psychological impediments to follow-up; cardiogenic shock; recent MI; and other major cardiac or medical problems likely to affect prognosis) resulted in an overall Ml rate of 86% in patients determined to be eligible. This represents the overall value of our criteria in selecting, before serial electrocardiographic or serum enzyme data are available, a population with a high prevalence of acute MI.