This study was undertaken to determine the effects on systemic fibrinolysis of hyperthermic isolated limb perfusion with recombinant tumor necrosis factor alpha (r-TNF-alpha) and melphalan, with or without pretreatment with recombinant IFN-gamma (r-IFN-gamma). Twenty patients were treated with r-TNF-alpha and melphalan; four patients, treated with melphalan only, served as controls. Of the twenty patients treated with both r-TNF-alpha and melphalan, eight received r-IFN-gamma for two days before the perfusion and as a bolus into the perfusion circuit. A significant leak of r-TNF-alpha from the perfusion circuit to the systemic circulation was observed in all r-TNF-alpha-treated patients (mean maximum TNF-alpha, 87,227 ng/liter versus 31 ng/liter in controls; P < 0.002). In these patients, but not in controls, there was an almost instantaneous rise in systemic tissue plasminogen activator activity (from 0.26 to 5.28 IU/ml in 90 min), causing activation of fibrinolysis. After a delay of 90 min, plasminogen activator inhibitor-1 (PAI-1) antigen rose to high levels in the r-TNF-alpha-treated group (mean maximum PAI-1, 1652 ng/ml versus 211 ng/ml in controls; P < 0.02), associated with a sharp decrease of tissue plasminogen activator activity and a slower decrease of plasminogen-antiplasminogen complexes (from 5.28 to 0.02 IU/ml in 2 h and from 1573 to 347 micrograms/liter in 22 h, respectively). No additional effect of IFN-gamma pretreatment on fibrinolysis could be demonstrated. These results suggest that in isolated limb perfusion with r-TNF-alpha and melphalan an initial activation of systemic fibrinolysis, induced by leakage of r-TNF-alpha from the perfusion circuit, is set off by a subsequent inhibition of the fibrinolytic system by PAI-1. This large increase in PAI-1 could place the patient at risk for deposition of microthrombi in the systemic circulation.