Five hundred and seventy physicians, researchers and clinicians (42% response) responded to a mailed questionnaire about the safety and nature of exercise testing conducted less than or equal to 4 weeks after acute myocardial infarction (AMI). Of 570 institutions, 193 reported that they routinely performed testing early after AMI and data were provided on 151,949 tests. A majority (111 or 58%) used a low-level testing protocol, 50 (26%) used symptom-limited testing and 32 (16%) used both types. Testing was routinely conducted less than or equal to 14 days after AMI by 147 (76%) respondents, whereas 46 (24%) tested 15 to 28 days after AMI. Thirty-three (17%) respondents used a standardized research protocol and 160 (83%) did not. There were 41 (0.03%) fatal, 141 (0.09%) major nonfatal and 2,124 (1.4%) other cardiac complications reported during testing. No difference in incidence of major complications was observed at centers using a clinical versus research protocol. Compared with clinic-based testing, hospital-based testing had an increased risk for all major (2.1) and nonfatal major complications (2.1). Although a symptom-limited protocol increased the overall risk for major cardiac complications by 1.9 times compared with a low-level protocol, the incidence of fatal complications during symptom-limited testing (0.03%) was quite low and this greater risk is of dubious clinical importance.