Understanding the clinical risks of intravenous thrombolytic therapy is critical to appropriate patient selection. The major risks can be classified into 5 major categories: intracranial hemorrhage, systemic hemorrhage, immunologic complications, hypotension, and myocardial rupture. Although theoretical concern exists about thromboembolic complications, they rarely occur. Although cardiac rhythm disturbances are somewhat more likely to occur at the time of reperfusion, the clinical significance of "reperfusion arrhythmias" is minimal. Intracranial hemorrhage, the most devastating complication, occurs in 0.2-1% of patients treated with thrombolytic therapy. Factors associated with incremental risk are now being identified from large clinical trials. Systemic hemorrhage is uncommon in patients without major vascular punctures and seldom leads to serious adverse outcomes. Immunologic complications--including anaphylaxis, which is rare, and immune complex disease, which is more common--occur only with streptokinase or agents with a streptokinase moiety, including anistreplase (anisoylated plasminogen--streptokinase activator complex, APSAC). Hypotension, which can be managed easily in most patients, is also observed much more frequently with streptokinase and anistreplase. Myocardial rupture is increasingly being recognized as a possible complication of late thrombolysis. A proper perspective on clinical risk can only be gained in the context of potential benefit of therapy. In many cases individual patients considered to be at highest risk for complications also stand to gain the most from treatment. Many of the questions raised by currently available data about bleeding risk are being addressed in the ongoing Global Utilization of t-PA and Streptokinase (GUSTO) Trial. A paradigm for considering this decision making problem is presented.