PURPOSE: The objective of this study was to investigate the outcome of revision total knee arthroplasty (TKA) in relation to the cause of index failure, the characteristics of the index procedure, and the elapsed time between index TKA and revision. METHODS: A retrospective review based on a prospective database was performed on 146 consecutive revision TKA's. Variables tested were the cause of index failure; the elapsed time between the index and revision procedure; patient age at time of revision; partial or total revision of the implants; the performance of a tibial tubercle osteotomy; the presence of radiolucent lines; postoperative patellar tracking; and coronal plane alignment. Outcomes were measured with the Knee Society Knee Score (KS), Function Score (FS), and X-ray evaluation. RESULTS: Mean KS improved from 27.6 (SD 21.6) to 71.5 (SD 24.2) after revision (P < 0.0001), mean FS from 27.5 (SD 22.7) to 53.3 (SD27.7), P < 0.0001. Overall survival rate was 90% at 5 years and 85% at 10 and 14 years. The cause of index failure had no significant influence on any of the outcome parameters. Significantly, better outcomes were noted for partial revisions and for revisions in older patients. Early revisions (<2 years) were mostly performed for infection and instability, whereas late revisions (>2 years) were mostly performed for polyethylene wear and loosening. The survival rate for late revisions was significantly better than for early revisions (P = 0.002). CONCLUSION: Revision TKA leads to a significant reduction in symptoms and improvement in function. The worst results can be expected for early revisions in young patients. Revision TKA is a demanding procedure with variable results and should therefore be performed by experienced surgeons. LEVEL OF EVIDENCE: Therapeutic study-Level IV.