Background The optimal heparin dose as an adjunct to fibrinolysis and its role in causing intracranial hemorrhage (ICH) is unclear. Methods We reviewed the heparin regimens and rates of ICH in 3 sets of recent fibrinolytic trials: (1) studies with accelerated recombinant tissue plasminogen activator (TPA, alteplase) plus intravenous heparin, in which the heparin regimen was changed during the course of the trial; (2) phase III trials with accelerated TPA plus intravenous heparin; and (3) trials of new single-bolus fibrinolytic agents. Results Lower rates of ICH were observed among studies of accelerated TPA that reduced the heparin dose mid-trial (TIMI 9A → 9B: 1.87% → 1.07%, GUSTO-IIa → IIb: 0.92% → 0.71%, TIMI 10B: 2.80% → 1.16%). Rates of ICH with accelerated TPA gradually increased from GUSTO-I (0.72%) in 1990 to 1993 to ASSENT-2 (0.94%) in 1997 to 1998. However, this trend was reversed in InTIME-II, which used the lowest heparin dose and most aggressive activated partial thromboplastin time monitoring and observed an ICH rate of 0.64% with accelerated TPA. Lower ICH rates were also observed when the heparin dose was reduced with single-bolus tenecteplase (TNK-TPA) and lanoteplase. Conclusions Nonrandomized comparisons with accelerated TPA suggest that lower doses of intravenous heparin are associated with lower rates of ICH. This observation also appears to apply to single-bolus TNK-TPA and novel plasminogen activator. A lower-dose, weight–adjusted heparin regimen (60 U/kg bolus; maximum, 4000 U; 12 U/kg per hour infusion; maximum, 1000 U/h) with earlier monitoring of activated partial thromboplastin time is currently recommended in the revised American College of Cardiology/American Heart Association myocardial infarction guidelines and should be used in clinical practice. (Am Heart J 2001;141:742-50.)