In a rabbit jugular vein thrombosis model, the thrombolytic properties of melanoma tissue-type plasminogen activator (mt-PA), recombinant t-PA (rt-PA), and a t-PA mutant that lacked the growth factor- and kringle one-domain (designated FK2P, as it contained the finger-, kringle two-, and protease-domain) were compared. Over the dose range tested (3.75 to 30 nmol/kg) a similar, dose-dependent, thrombolytic efficacy was found for the 3 enzymes when given as bolus injections. When given as a 90-min infusion, FK2P was superior to rt-PA at the tested dose of 15 nmol/kg. The effects on systemic plasma fibrinogen, plasminogen and alpha 2-antiplasmin consumption were much larger for rt-PA and FK2P than for mt-PA. The analogue FK2P was cleared more slowly than mt-PA or rt-PA, despite a similar initial alpha half-life: the mean residence time of FK2P was about 10 times as large as that of mt-PA or rt-PA. A comparison with data obtained in vitro suggested that in vivo thrombolytic performance cannot easily be predicted from in vitro data. The efficacy of a thrombolytic agent will also depend on its clearance rate, and may also, within a chosen model, depend on the way it is administered.