OBJECTIVE--To assess the clinical ability of general practitioners to decide to give thrombolytic therapy to patients with suspected myocardial infarction and to assess the contribution of the electrocardiograph (ECG) to this decision-making process. SETTING--7 practices on the North side of Glasgow and the coronary care unit of Stobhill General Hospital. SUBJECTS--137 patients presenting with chest pain who required direct admission to the coronary care unit. MAIN OUTCOME MEASURES--Agreement between the general practitioner's clinical decision to give thrombolytic therapy with or without reference to the ECG and the prescription of thrombolytic therapy in the coronary care unit. RESULTS--The predictive accuracy of the general practitioner's assessment of the necessity for thrombolytic therapy was 71.5%. The ECG had no impact on the accuracy of this decision and there were problems with the recording and interpretation of the ECG. Clinical decision making was altered in six cases by the ECG--wrongly in four. CONCLUSION--The diagnostic accuracy among general practitioners would result in some patients who did not have acute myocardial infarction being given thrombolytic therapy. In this study the ECG did not contribute towards diagnostic accuracy. Substantial improvement in both the recording and interpretation of ECGs is needed before thrombolytic agents can be routinely prescribed at home.