Computer-assisted surgery (CAS) systems improve alignment accuracy in total knee arthroplasty (TKA) but have not been widely implemented. Eighty knees underwent TKA using an accelerometer-based, portable navigation device (KneeAlign 2; OrthAlign Inc, Aliso Viejo, California), and the radiographic results were compared with 80 knees performed using a large-console, imageless CAS system (AchieveCAS; Smith and Nephew, Memphis, Tennessee). In the KneeAlign 2 cohort, 92.5% of patients had an alignment within 3° of a neutral mechanical axis (vs 86.3% with AchieveCAS, P < .01), 96.2% had a tibial component alignment within 2° of perpendicular to tibial mechanical axis (vs 97.5% with AchieveCAS, P = .8), and 94.9% had a femoral component alignment within 2° of perpendicular to the femoral mechanical axis (vs 92.5% with AchieveCAS, P < .01). The mean tourniquet time in the KneeAlign 2 cohort was 48.1 ± 10.2 minutes vs 54.1 ± 10.5 minutes in the AchieveCAS cohort (P < .01). Accelerometer-based, portable navigation is as accurate as large-console, imageless CAS systems in TKA.