Objective To explore the clinical efficacy of single unicompartmental knee arthroplasty (UKA) and total knee arthroplasty (TKA) for the treatment of knee osteoarthritis. Methods A total of 21 patients who underwent TKA and 15 who underwent UKA were randomly recruited. Biomechanical tests were performed before surgery and at 6th and 12th month after surgery. A Vicon infrared motion capture system and Kistler three-dimensional force plate were used to simultaneously acquire the kinematic and kinetic data of the patients during stair walking. Results During stair ascent, the peak knee flexion moment in the TKA group was significantly lower than that in the UKA group; the time to peak knee flexion/adduction moment, knee flexion moment impulse, and load rate of the peak knee adduction moment in the UKA group were significantly lower than those in the UKA group. During stair descent, the peak knee extension power in the UKA group was significantly lower before surgery and at 6th month after surgery; the load rate of the peak vertical ground reaction force was significantly higher before surgery and the peak knee extension moment was significantly greater at 6th month after surgery; at 12th month after surgery, there was no significant difference in the biomechanical characteristics during stair ascent and descent. Conclusions The TKA and UKA groups showed similar knee joint function after surgery; however, compared with the UKA group, the TKA group may adopt a different lower extremity biomechanical pattern. The UKA group showed better quadriceps control after surgery and improved postural control during stair descent, whereas the TKA group adopted a conservative stair gait strategy to reduce the knee load. Compared with the peak moment, the time to peak moment and load rate of the peak moment were more sensitive indicators for determining the difference in the knee load.