We empirically developed a logistic function to predict acute coronary heart disease (ACHD) and then tested it to determine its usefulness to emergency room physicians in diagnosis and admission of patients with suspected ACHD to the coronary care unit. The function was based on nine clinical, historical and electrocardiographic variables from a set of 105 variables collected on 643 patients with suspected ACHD. In the second phase of the study, we provide the probabilities generated by the function to emergency room physicians during alternate months as a supplement to existing diagnostic modalities. Use of the probability of ACHD (401 patients with probabilities versus 455 patients without probabilities) resulted in statistically significant improvement in diagnostic rates and reduction in the number of inappropriate admissions to the coronary care unit.