Previous studies of biventricular (BV) pacing for treatment of heart failure (HF) patients with left bundle branch block (LBBB) evaluated responders to BV pacing with acute transvenous left ventricular (LV) pacing and arterial pulse pressure (PP). The aim of this study was to assess transoesophageal LV pacing in evaluation of the haemodynamic response with a view to upgrading responders from permanent right ventricular (RV) pacing to BV pacing.Ten HF patients (age 62+/-8 years; one female, nine males) in NYHA III, LV ejection fraction 24+/-9% and permanent RV pacing by means of an implanted pacemaker or ICD were tested using transoesophageal LV pacing and PP. Permanently RV-paced HF patients were analysed with transoesophageal atrial sensed LV pacing in VAT mode with a different AV delay (n = 6) and with transoesophageal LV pacing in V00 mode during atrial fibrillation (n = 4). In five responders, PP was higher during transoesophageal LV pacing than PP during RV pacing (74+/-42 versus 57+/-31 mmHg, P = 0.015). Responders were upgraded by means of an LV lead via the coronary sinus in the posterior (n = 1) or posterolateral (n = 4) walls and after attaining a high LV pacing threshold with an epicardial LV lead on the anterior (n = 1) or anterolateral (n = 1) walls. NYHA class improved from 3 to 2+/-0.3 (P = 0.003) during 204+/-120 days follow-up and cardiac output increased from 4.4+/-1.5 to 5.6+/-1.7 l/min (P = 0.027) when comparing BV pacing and optimal AV delay with RV pacing. In five nonresponders, PP was not higher during transoesophageal LV pacing than during RV pacing.Transoesophageal LV pacing may be a useful technique to detect responders to BV pacing in permanently RV-paced HF patients.