Computer-assisted surgery (CAS) has been advocated as a means to improve limb and prosthesis alignment and assist in ligament balancing in total knee arthroplasty (TKA). Thus, we sought to examine alignment outcomes in CAS vs conventional TKA. A systematic review of literature from 1990 to 2007 was performed. Direct comparison of alignment outcomes was performed using random effects meta-analyses. Twenty-nine studies of CAS vs conventional TKA were identified, and included mechanical axis malalignment of greater than 3 degrees occurred in 9.0% of CAS vs 31.8% of conventional TKA patients. The risk of greater than 3 degrees malalignment was significantly less with CAS than conventional techniques for mechanical axis and frontal plane femoral and tibial component alignment. Tibial and femoral slope both showed statistical significance in favor of CAS at greater than 2 degrees malalignment. Meta-analysis of alignment outcomes for CAS vs conventional TKA indicates significant improvement in component orientation and mechanical axis when CAS is used.