Purpose: The present study was designed to compare the long-term outcome and complications of cemented versus screw-retained implant restorations in partially edentulous patients. Materials and Methods: Consecutive patients with bilateral partial posterior edentulism comprised the study group. Implants were placed, and cemented or screw-retained restorations were randomly assigned to the patients in a split-mouth design. Follow-up (up to 15 years) examinations were performed every 6 months in the first year and every 12 months in subsequent years. The following parameters were evaluated and recorded at each recall appointment: ceramic fracture, abutment screw loosening, metal frame fracture, Gingival Index, and marginal bone loss. Results: Thirty-eight patients were treated with 221 implants to support partial prostheses. No implants during the follow-up period (mean follow-up, 66 ± 47 months for screw-retained restorations [range, 18 to 180 months] and 61 ± 40 months for cemented restorations [range, 18 to 159 months]). Ceramic fracture occurred significantly more frequently (P .001) in screw-retained (38% ± 0.3%) than in cemented (4% ± 0.1%) restorations. Abutment screw loosening occurred statistically significantly more often (P = .001) in screw-retained (32% ± 0.3%) than in cement-retained (9% ± 0.2%) restorations. There were no metal frame fractures in either type of restoration. The mean Gingival Index scores were statistically significantly higher (P .001) for screw-retained (0.48 ± 0.5) than for cemented (0.09 ± 0.3) restorations. The mean marginal bone loss was statistically significantly higher (P .001) for screw-retained (1.4 ± 0.6 mm) than for cemented (0.69 ± 0.5 mm) restorations. Conclusion: The long-term outcome of cemented implant-supported restorations was superior to that of screw-retained restorations, both clinically and biologically. Palabras clave: cementation, implant-supported restoration, partial edentulism, screw retention