We determined the incidence of persistent back pain (PBP) after non-obstetrical spinal anaesthesia (SPA) and investigated factors predisposing to such pain in a prospective 1 yr follow-up study in 245 patients undergoing elective general or trauma surgery (218 patients undergoing single SPA, 27 undergoing two to six SPAs). All patients received a first questionnaire 3 months after the last SPA, and those reporting PBP after 3 months were sent a second questionnaire I year after the operation. Variables were PBP before and within 5 days, at 3 months and I year after SPA, patient satisfaction with SPA, patient characteristics and technical data. Statistical analysis was by contingency tables with Fisher's exact test and an unpaired t-test with logistic regression (P < 0.001 after Bonferroni correction was taken as significant). The response rate in patients who had a single SPA was 56% (122/218). Twenty-three of these 122 patients (18.9%) complained of back pain before SPA compared with 12/122 (10.7%, P = 0.0015) within 5 days after SPA. After 3 months, 15/122 patients (12.3%) reported PBP with 14 complaining of PBP before SPA (P < 0.0001), corresponding to an incidence of new PBP of 1/122 (0.8%). Multiple logistic regression revealed that pre-existing back pain was the only variable associated with PBP after 3 months (P < 0.0001). Patient characteristics and technical factors were not associated with PBP. Nine of the 15 patients with PBP after 3 months returned the second questionnaire: four still reported PBP (three of these had suffered from PBP before SPA). Despite PBP after 3 months, 13/15 patients said they would opt for SPA again. The response rate and results in patients who had had multiple SPAs were similiar to those who had had a single SPA.