Objective: To evaluate the efficacy of Tuina (Chinese therapeutic massage) manipulation plus horse-riding squat exercise in treating knee osteoarthritis (KOA) and optimize the combining protocol. Methods: Based on a 2×2 factorial design, 120 eligible KOA patients were randomized into a manipulation group (group A1B2), a manipulation plus horse-riding squat group (group A1B1), a sitting knee-adjustment group (group A2B2 group), and a sitting knee-adjustment plus horse-riding squat group (group A2B1), with 30 cases in each group. The intervention was conducted three times a week, lasting for four weeks. The Western Ontario and McMaster Universities osteoarthritis index (WOMAC) was taken as the major measure for efficacy evaluation (including three component scores, pain, stiffness, and daily function, and total score). Results: The three component scores (pain, stiffness, and daily function) and the total score of WOMAC showed significant differences after the intervention in the four groups (P<0.05). There were significant inter-group differences in the WOMAC stiffness score amongst the four groups after the intervention (P<0.05). In group A1B1, the step length, stride, walking speed, and knee joint flexion angle changed significantly after treatment (P<0.05). After the intervention, the step length changed significantly in group A1B2 (P<0.05), and the walking speed changed significantly in group A2B1 (P<0.05). There were no significant differences in the step length, stride, walking speed, or knee joint flexion angle among the four groups (P>0.05). The extensor peak torque at 180 °/s changed significantly in group A1B2 after treatment (P<0.05). Neither the intra-group nor the inter-group comparisons of the four groups revealed significant differences in the other isokinetic muscle strength parameters (P>0.05). The main effect of manipulation showed significant in affecting the WOMAC pain and total scores (P<0.05). The main effect of horse-riding squat exercise showed significant in affecting the WOMAC pain and stiffness scores (P<0.05). Conclusion: The four treatment protocols all can improve the symptoms of KOA, for instance, relieving pain and stiffness, and enhancing daily function. Group A2B1 produces the most eminent effect in relieving joint stiffness. The main effects of both manipulation and horse-riding squat exercise are significant in reducing pain. Besides, the main effect of horse-riding squat exercise is significant in relieving joint stiffness.