This paper describes a preliminary experiment-conducted jointly by 2 centers-of permanent left ventricular pacing using leads inserted by the transvenous route and through the coronary sinus into the cardiac veins of the left ventricle free wall. The aim was to obtain permanent biventricular pacing in a totally endocavitary configuration in patients with severe LV dysfunction and drug-refractory heart failure. Two types of leads were used: nonspecific unipolar leads at the beginning of the experiment, followed by leads specifically designed to be used in the coronary sinus in a second step. The electrode could be fitted in an adequate location in 35 of the 47 patients (75.4%), with a 1.15 +/- 0.7 V acute pacing threshold and 11.8 +/- 5.7 mV R wave amplitude. The success rate was significantly higher with the specific electrodes (81.8% vs 53.3%, p < 0.001). The pacing and sensing thresholds upon implantation were not influenced by the type of lead or by the localization of the cardiac vein that was catheterized (great cardiac vein, lateral vein, postero-lateral or posterior vein, mid cardiac vein). In contrast, the pacing threshold was significantly lower (0.8 +/- 0.2 vs 1.8 +/- 0.8 V; p = 0.002) and the R wave amplitude tended to be greater (13.1 +/- 4.5 mV vs 9.3 +/- 6.5 mV; p = 0.07) when the tip electrode could be inserted distally into the vein, by comparison with a proximal site near the ostium. At the end of follow-up (10.2 +/- 8.7 months), 34 out of the 35 leads were still fully functional, with a chronic pacing threshold of 1.8 +/- 0.7 V and a R wave amplitude of 10.7 +/- 6 mV. To conclude, permanent LV pacing via the transvenous route is possible in most patients, with excellent safety and long-term results.