Adverse outcome data from two insurance companies were retrospectively studied to determine whether a constellation of clinical circumstances, data-gathering behaviors, or physician variables were common to cases of missed acute myocardial infarction (AMI) and, if so, to formulate quality assurance recommendations to decrease future occurrences of misdiagnosis. We studied AMI because missing this diagnosis accounts for the highest dollar losses in emergency department malpractice cases. Our study group consisted of 65 patients with undiagnosed AMI seen in EDs between 1982 and 1986. Univariate differences between undiagnosed cases and correctly diagnosed concurrent controls were analyzed using Student's t test and chi 2 analysis. Insurance losses for our cases averaged $113,806 +/- $178,330 (SD). Compared with concurrent controls, study patients were significantly younger, presented more atypically, and had fewer ECGs that were diagnostic of AMI. Undiagnosed patients were evaluated by physicians who documented less detailed histories, misread more ECGs, had less ED experience, and admitted fewer patients to the hospital. Preventive strategies are outlined.