Objective To compare the therapeutic effect of conventional discectomy, posterior lumbar interbody fusion (PLIF), and transforaminal lumbar interbody fusion (TLIF) on the recurrent lumbar disc protrusion (RLDP). Methods From January 2000 to January 2008, 65 patients with RLDP underwent different surgical procedures, namely conventional discectomy (group A, 25 cases), PLIF (group B, 22 cases), and TLIF (group C, 18 cases). There were 44 males and 21 females aged 26-65 years old (average 41 years old). All the patients were single-level protrusion, including 33 cases at the L4, 5 level and 32 cases at the L5, S1 level. The primary procedure included laminectomy discectomy in 39 patients, unilateral hemilaminectomy discectomy in 15 patients, and bilateral laminectomy and total laminectomy discectomy in 11patients. The recurrent time to the primary operation was 13-110 months (average 64 months). The location of recurrent disc protrusion was at the ipsilateral side in 47 cases and the contralateral side in 18 cases. No significant differences among three groups were evident in terms of basel ine data (P gt; 0.05). Results The incision all healed by first intention. The incidence of perioperative compl ication in group A (24.0%) and group B (22.3%) was significantly higher than that of group C (5.6%) (P lt; 0.05), and there was no significant difference between group A and group B (P gt; 0.05). The operation time and bleed loss during operation of group B were obviously higher than that of group A and group C (P lt; 0.05), and there was no significant difference between group A and group C (P gt; 0.05). There were no significant differences among three groups in terms of the length of hospital ization (P gt; 0.05). Six-one patients were followed up for 12-36 months (average 20 months). At 1 week after operation, the satisfied rate of patients was 84.0% in group A, 81.8% in group B, and 88.9% in group C (P gt; 0.05). All the patients in group B and group C achieved fusion uneventfully. There were no significant differences among three groups in terms of visual analogue scale (VAS) and Oswestry disabil ity index (ODI) when compared the preoperative value with the final follow-up value (P gt; 0.05). There was significant difference within group A, B, and C in terms of VAS and ODI when compared the preoperative value with the final follow-up value (P lt; 0.05), but there were no significant differences among three groups in the improvement rate (P gt; 0.05). The intervertebral space grading method proposed by Roberts et al. was adopted to evaluate the intervertebral space height (ISH), the preoperative value was 2.04 ± 0.93 in group A, 2.18 ± 0.91 in group B, and 2.11 ± 0.90 in group C, andat the final follow-up, the value was 2.64 ± 0.58 in group A, 1.05 ± 0.59 in group B, and 1.06 ± 0.42 in group C. There were significant differences among three groups in the ISH when compared the properative value with the final follow-up value (P lt; 0.05). Conclusion All of the three surgical procedures are effective for RLDP, but conventional discectomy and PLIF have more compl ications than TLIF. Conventional discectomy may result in the further narrow of the intervertebral space and the occurrence of segment instabil ity, whereas TLIF is safer, more effective, and one of the ideal methods to treat RLDP.