Seventy two patients undergoing lumbar chemonucleolysis (LCN) involving one (49 cases) or two (23 cases) discs underwent surgery after failure of the procedure. In 5 cases the disc prolapse responsible for the recurrence of pain was at a different site from that of the LCN. Analysis of true failures in the light of operative findings showed that the presence of a migrated disc sequestrum at a distance from the space injected was rare (7% of cases). By contrast, the discovery of a true sub-ligamentous disc prolapse was common (32%). In 1/3 of cases these prolapses contained actual intra-discal sequestra within a scanty amount of basic substance, as if the enzyme had destroyed the latter but had no action upon the former. Simple disc protrusions were discovered in 36% of cases, combined in one case out of five with stenosis of a lateral recess. Finally, in 25% of cases the disc explored was flat, combined in 1 case out of 3 with a considerable inflammatory reaction. The percentage of good and very good operative results was 60.7% overall, rising to 70% taking into consideration only those patients undergoing LCN at one level only. The authors feel that no longer than a month should be allowed to pass before deciding to operate after failure of LCN.