Purpose: To compare the marginal vertical misfit between implant-supported frameworks fabricated using CAD/CAM systems and the conventional technique (lost-wax casting). Materials and Methods: This review was performed according to PRISMA criteria and registered on PROSPERO (CRD42017055685). An electronic search was performed independently by two examiners in the MEDLINE (Pubmed), Embase, Web of Science, and Cochrane Library databases to find studies published up to April 2018. Results: The database search yielded 507 references. After removing duplicate references, 384 studies remained. Eleven in vitro studies were selected according to the eligibility criteria (inter-reader κ = 0.88). Nine different CAD/CAM systems were used to fabricate 172 frameworks of different materials, including zirconia, monolithic lithium disilicate, and metallic alloys. Subgroup analyses were performed for different types and retention systems of the frameworks. In the general analysis, marginal misfit observed with the CAD/CAM systems was lower than with the conventional method (P = .003), as was observed in the subgroup analysis for single-unit frameworks (P .00001). For fixed (P = .89), cemented (P = .60), and screwed (P = .18) frameworks, no significant difference was observed between the evaluated techniques. Conclusion: The CAD/CAM systems showed improved marginal fit over the conventional lost-wax casting technique for fabricating single-unit frameworks; however, in the subgroup analyses, no difference was observed for the fixed implant-supported type or for the retention systems evaluated.