Fibrinolytic drugs and aspirin, compared with placebo, have reduced the 35-day mortality of patients with acute myocardial infarction from approximately 12% to about 8%. The mortality reduction with active treatment is most evident in high risk groups of patients. Moreover, mortality with a fully patent infarct-related artery at 90 min from the start of thrombolytic treatment is half that found with an occluded artery. Yet more than 50% of acute infarct patients receiving thrombolysis fail to achieve complete early coronary patency or develop re-occlusion. Although thrombolytic strategies are continually evolving to try to further reduce early mortality and increase coronary reperfusion rates, new regimens, when tested in unselected patients against successful active treatments, such as streptokinase and aspirin, are likely to show no, or only small, average benefits, even when applied to tens of thousands of patients. In contrast, the effect of a new thrombolytic regimen is likely to be most evident is selected patients showing large areas of potentially salvageable ischaemic myocardium, as these patients have a higher absolute risk of premature death than patients with smaller infarcts and therefore should gain greater benefit from successful reperfusion. Since the effect of the new treatment, if present, would be greatest in this group, a smaller number of patients would be needed to show an effect; moreover, the risk of the new treatment would be justified against the larger potential benefit to be gained. We propose to target future thrombolytic regimens, in the first instance, at homogenous groups of high-risk patients; for example, those with large areas of potentially salvageable myocardium. This approach seems a rational and cost-effective way to invest limited amounts of medical resources and offers patients the greatest potential benefit of the new therapy, while restricting possible complications to a smaller number. Continued clinical research aimed at identifying possibly different causes of coronary occlusion and the reasons for worse outcomes is also essential to assign the most appropriate treatments to selected groups of patients.