Seventy-seven patients with chronic, refractory, low back and radicular pain underwent implantation of a spinal cord stimulator between 1984 and 1992. Most patients had failed back surgery syndrome. In every case, an epidural quadripolar "Resume" electrode was implanted surgically. Results were evaluated after three months then after six to 98 months (mean follow-up 42 months). Long-term efficacy was good in 63.6% of cases, fair in 22%, and poor in 6.5%; treatment failure occurred in 7.9% of cases. Adverse events included one case of meningitis, two cases of local infection, and one case each of cerebrospinal fluid fistula and necrosis of the skin overlying the stimulator. The main causes of treatment failure were complications, inappropriate patient selection, and the escape phenomenon. The results of this study demonstrate that spinal cord stimulation is effective for the treatment of chronic low back and radicular pain in carefully selected patients; scrupulous application of restrictive selection criteria is essential to the success of the method.