In recent trials, patients with myocardial infarction who received either recombinant tissue-type plasminogen activator (rt-PA) or streptokinase showed essentially no difference in the amount of myocardial salvage, in mortality reduction, or in the incidence of bleeding complications. These findings thus failed to fulfill the expectation that rt-PA would be twice as effective as streptokinase as a thrombolytic agent. The basis for this mistaken prediction was an unfortunate overemphasis on an inadequate surrogate endpoint, namely, the patency or reperfusion rate at 90 minutes after the start of therapy. Using the 90-minute patency or reperfusion rate as an endpoint has several serious limitations. First, it is an observation made at only one point in time during a dynamic process that may change even during the infusion proper. Second, a single view at 90 minutes completely disregards the possibility of subsequent reocclusion which often occurs within 1 hour after treatment. Third, an image at 90 minutes is more a reflection of the speed of thrombolysis than of whether lysis will eventually occur; the pace of clot lysis depends on both the agent used and the age of the thrombus. Fourth, lysis at 90 minutes is of minimal relevance for myocardial salvage unless observed within the time frame when infarction size can be limited significantly, which is generally less than 4 hours between symptom onset and the time that reperfusion is accomplished. Fifth, a stable state of vessel patency is meaningful for mortality reduction even if stabilization occurs after completion of the infarction. Such "late," but lasting, patency is a critical component of the "open vessel" principle and explains, at least in part, the survival benefit that accrues to patients treated even 24 hours after the onset of symptoms. There is currently no evidence that rt-PA has a more beneficial effect on survival or function than does streptokinase or any other plasminogen activator used in treating acute myocardial infarction; nor is there any evidence that patients who receive rt-PA therapy show a decreased incidence of bleeding complications compared with those who receive streptokinase, despite the relative fibrinogen-sparing attribute of rt-PA. Given the poor predictive value of the 90-minute angiogram for ultimate clinical advantage of one agent over another, studies that are limited to this endpoint are of marginal use in evaluating treatment regimens used in mortality studies. The best evidence to date indicates that streptokinase and rt-PA are of equivalent value for survival after acute myocardial infarction, a conclusion that can be justifiably challenged only with a valid mortality study.