The benefits of exercise are well established for psychological well-being, but it is unknown whether one type of exercise can be recommended over another for psychological well-being in knee osteoarthritis (OA). This study aimed to compare the effectiveness of different types of exercise on psychological well-being in people with knee OA. Five databases including MEDLINE, EMBASE, Cochrane Register of Controlled Trials, CINAHL and PEDro) were searched from inception until April 2021. Randomised controlled trials comparing the effect of exercise with a non-exercise control or another type of exercise on overall mental health, depressive symptoms or anxiety in people with knee OA. Trial data were extracted independently by two researchers. Network meta-analyses using random-effect models were conducted. The Grading of Recommendations, Assessment, Development and Evaluation framework was used to appraise evidence certainty. Strengthening exercise was more beneficial for overall mental health on the Short Form survey compared to aerobic (mean difference 15.88, 95%CI [6.77, 24.99]), mixed (12.77, 95%CI [5.12, 20.42]) and mind-body (12.51, [4.25, 20.77]) exercise based on 13 trials (n = 796). Strengthening exercise (standardised mean difference 6.81, [5.03, 8.58]), and mixed exercise (6.64, [4.88, 8.40]),) were more beneficial for depressive symptoms than stretching exercise based on 6 trials (n = 627). No differences were observed for anxiety based on 4 trials (n = 557). Certainty of the evidence ranged from very low to low. The available evidence supports the recommendation of strengthening exercise as the most beneficial for overall mental health, and strengthening exercise or mixed exercise over stretching exercise for depressive symptoms. No exercise recommendations can be made for anxiety.