The place of fibrinolysis in the treatment of mechanical valvular prostheses is still much discussed. The aim of this work is to define the role of transoesophageal echocardiography in risk stratification. This monocentric study draws on 49 cases of thrombolysis preceded by transoesophageal echocardiography (average age 62.1 years, 37 mitral, 11 aortic, 1 tricuspid, 1 mitro-tricuspid). There were 41 obstructive thromboses (OT) and 8 non-obstructive thromboses (NOT). Clinical events and the effectiveness of fibrinolysis were studied as a function of the obstructive or non-obstructive character of the thrombosis and the size of the thrombus < 10 mm (n = 33) or > or = 10 mm (n = 16). Complete success was observed in 34 patients (69.4%). Follow up revealed 2 early cerebral haemorrhages (4.1%) of which one was in the NOT group, and six systemic emboli (12.2%) of which one was in the NOT group. There was a relationship between the size of the thrombus and embolus at the limit of significance in favour of an increased risk of embolus for a voluminous thrombus. Furthermore, the mobility of the thrombi went in hand with an increased rate of systemic emboli (p < 0.01). The rate of failure of fibrinolysis and/or complications correlated with the size of thrombus (complete success in 88% of the < 10 mm thrombus group, versus 35% in the > or = 10 mm; p < 0.01). This work underlines the significance of trans-oesophageal echocardiography in the therapeutic choice for valvular prosthesis thrombosis and suggests that the existence of a voluminous thrombus especially if mobile is a contra-indication for fibrinolysis.