To evaluate the long-term effectiveness of a single cervical epidural steroid injection (CESI) performed with or without morphine, 24 patients, without need of surgery, but suffering for more than 12 months from cervical radicular pain, were included in a prospective and randomised study. The cervical epidural space was injected (C7-D1; 18-ga needle) with an increasing volume (10 ml maximum) of isotonic saline solution to exacerbate the patient's radicular pain. The patients were then randomly allocated to 2 groups: the steroid group (group S, n = 14) received an equivalent volume of 0.5% lidocaine plus triamcinolone acetonide (10 mg/ml) and the steroid plus morphine group (group S + M, n = 10) received the same combination plus 2.5 mg of morphine sulphate. Pain relief was assessed as the percentage of pain decrease on a visual analogue scale on day 1 and at months 1, 3, 6, 8 and 12 after CESI, up to 48 months. Anthropometric data between the 2 groups were similar. The mean volume injected in the epidural space was: 6.6 +/- 2.1 and 6.3 +/- 1.9 ml in groups S and S + M, respectively, and this volume exacerbated pain in 21 of 24 patients. Despite observing a better transient improvement the day after CESI in the S + M group, long-term results did not differ. The success rate was 78.5% in group S and 80% in group S + M providing pain relief of 86.8 +/- 14.7% and 86.9 +/- 17.9%, respectively. Pain relief remained stable with time (mean follow-up: 43 +/- 18.1 months).(ABSTRACT TRUNCATED AT 250 WORDS)