Two hundred and twenty-eight Japanese women, aged 45-69 years, with osteoarthritis of the knee (knee OA) and a body mass index of greater than 26.4 self-selected one of the following interventions: (1) a nonsteroidal anti-inflammatory drug (NSAID) alone (NA group; n = 52), (2) the NSAID with nonweight-bearing exercises (NE; n = 49), (3) the NSAID combined with walking (NW; n = 35), (4) an energy restriction diet plus the NSAID (ND; n = 29), (5) the diet combined with the NSAID and strengthening exercises (NDE; n = 37), or (6) the diet combined with the NSAID and walking (NDW; n = 26). Body weight and lower extremity lean body mass (L-LBM), assessed with segmental bioelectrical impedance, were measured at the outset and at 8 weeks in the six therapy groups. Interval changes in L-LBM and L-LBM per body weight (L-LBM/weight) were compared among the six groups at the conclusion. The Lequesne index can index of severity of osteoarthritis of the knee at the final assessment, compared with that at the initial assessment, was significantly decreased in the NE group, but not in the NW or the NA groups (P = 0.023). There was a significant increase in the L-LBM/weight in the NE group, but not in the NA or NW groups (P = 0.002 between NE and NA; P = 0.019 between NE and NW). There was a significant reduction in the Lequesne index in the NDE group in comparison with the ND and NDW groups (P = 0.0001 between NDE and ND; P = 0.0001 between NDE and NDW). There was a significant increase in L-LBM/weight in the NDE group, but not in the NDW or the ND groups (P = 0.011). Adjunctive therapy with nonweight-bearing lower extremity exercises appears to be more efficacious for prevention of lower extremity loss of muscle mass than adjunctive therapy with walking for obese women with knee OA.