Fifty patients with chronic resistant cervicobrachialgia were randomly divided into two groups. Twenty-five patients (group A) were treated with cervical epidural steroid/lidocaine injections and 17 patients (group B) were treated with steroid/lidocaine injections into the posterior neck muscles. Another eight patients from group B were excluded from the study because they had started the process of litigation of insurance claims and their subjective analysis of pain relief might therefore not be trustworthy. One to three injections were administered at 2-week intervals according to the clinical response. All patients continued their various pre-study treatments: non-steroidal anti-inflammatory drugs, non-opioid analgesics and physiotherapy. Pain relief was evaluated by the visual analogue scale 1 week after the last injection and then 1 year later. One week after the last injection we rated pain relief as very good and good in 76% of the patients in group A, as compared to 35.5% of the patients in group B. One year after the treatment 68% of the group A patients still had very good and good pain relief, whereas only 11.8% of group B patients reported this degree of pain relief. These differences were statistically significant. We failed to achieve significant improvement of tendon reflexes or of sensory loss in both groups, but the increase in the range of motion, the fraction of patients who were able to decrease their daily dose of analgesics, and recovery of the capacity for work were significantly better in group A. We encountered no complications in either group of patients.(ABSTRACT TRUNCATED AT 250 WORDS)