OBJECTIVE: To determine the effect of spinal cord stimulation (SCS) on limb survival in patients with non-reconstructable critical leg ischaemia, and the value of patient selection on the basis of transcutaneous oxygen pressure (TcpO2) measurements and trial screening. DESIGN: A prospective, controlled, European multicentre study. METHODS: Non-reconstructable patients with stable critical leg ischaemia were divided into three groups. The SCS-Match group comprised patients with a baseline forefoot TcpO2 of < 30 mmHg and both sufficient pain relief and sufficient paraesthesia coverage (> 75%) after a test stimulation period of at least 72 h. If baseline TcpO2 was < 10 mmHg, the TcpO2 should have exceeded 20 mmHg after test stimulation. The SCS-Match group was compared with patients not meeting these criteria, who were treated either with SCS (SCS-No-Match) or without SCS (No-SCS). RESULTS: At baseline, the mean (+/- SD) supine TcpO2 was 14.9 +/- 8.3 mmHg in the SCS-Match group (n = 41), 11.3 +/- 13.3 mmHg in the SCS-No-Match group (n = 32) and 15.3 +/- 17.1 mmHg in the No-SCS group (n = 39). In the SCS-Match group a significant improvement in pain relief (p < 0.005) and TcpO2 (p < 0.001) was seen. After 12 months, cumulative limb survival of patients treated with SCS was significantly better than that of patients not treated with SCS (p < 0.03), and limb survival in the SCS-Match group was significantly higher (p < 0.03) than that in the SCS-No-Match and No-SCS groups (78, 55 and 45%, respectively). CONCLUSION: SCS treatment of non-reconstructable critical leg ischaemia provides a significantly better limb survival rate compared with conservative treatment. Patient selection based on TcpO2 and the results of trial screening further increase the probability of limb survival after SCS therapy.