It has been widely observed that the outcome after repeat lumbar surgery is rarely comparable to that of primary surgery. In particular, the results of repeat surgery for lumbar spinal stenosis (LSS) have not been favourable. We used a matched-pair format in an attempt to decrease the confounding factors so as to determine as exactly as possible the effect of prior back surgery on the LSS patients surgical outcome. The matching criteria were sex, age, myelographic findings, major symptom, and duration of symptoms. From one group of 251 patients without prior back surgery (SO patients) and another of fifty-three patients with one preceding back operation (RS patients), forty-one similar matched patients pairs (one SO and one RS-patient) were formed. There were 8 female and 33 male pairs. The mean age of the SO patients was 51.6 and of the RS patient 51.4 years, and the mean follow-up time was 4.6 and 4.4 years. The assessment of outcome was based on a subjective disability questionnaire. The SO patients fared significantly better than the RS patients (32.1 versus 41.3, P = 0.026). A short time interval between operations in the RS patients had a worsening effect on outcome, but this trend was not significant. We concluded that one preceding back operation had a worsening effect on the outcome of patients operated on for LSS. As a whole, the results of RS patients were unfavourable. The proper time for achieving good surgical results in LSS patients is the initial operation.