BACKGROUND: Sarcopenic obesity, associated with greater risk of cardiovascular disease (CVD) and mortality in rheumatoid arthritis (RA), may be related to dysregulated muscle remodeling. To determine whether exercise training could improve remodeling, we measured changes in inter-relationships of plasma galectin-3, skeletal muscle cytokines, and muscle myostatin in patients with RA and prediabetes before and after a high-intensity interval training (HIIT) program. METHODS: Previously sedentary persons with either RA (n = 12) or prediabetes (n = 9) completed a 10-week supervised HIIT program. At baseline and after training, participants underwent body composition (Bod Pod®) and cardiopulmonary exercise testing, plasma collection, and vastus lateralis biopsies. Plasma galectin-3, muscle cytokines, muscle interleukin-1 beta (mIL-1β), mIL-6, mIL-8, muscle tumor necrosis factor-alpha (mTNF-α), mIL-10, and muscle myostatin were measured via enzyme-linked immunosorbent assays. An independent cohort of patients with RA (n = 47) and age-, gender-, and body mass index (BMI)-matched non-RA controls (n = 23) were used for additional analyses of galectin-3 inter-relationships. RESULTS: Exercise training did not reduce mean concentration of galectin-3, muscle cytokines, or muscle myostatin in persons with either RA or prediabetes. However, training-induced alterations varied among individuals and were associated with cardiorespiratory fitness and body composition changes. Improved cardiorespiratory fitness (increased absolute peak maximal oxygen consumption, or VO2) correlated with reductions in galectin-3 (r = -0.57, P = 0.05 in RA; r = -0.48, P = 0.23 in prediabetes). Training-induced improvements in body composition were related to reductions in muscle IL-6 and TNF-α (r < -0.60 and P <0.05 for all). However, the association between increased lean mass and decreased muscle IL-6 association was stronger in prediabetes compared with RA (Fisher r-to-z P = 0.0004); in prediabetes but not RA, lean mass increases occurred in conjunction with reductions in muscle myostatin (r = -0.92; P <0.05; Fisher r-to-z P = 0.026). Subjects who received TNF inhibitors (n = 4) or hydroxychloroquine (n = 4) did not improve body composition with exercise training. CONCLUSION: Exercise responses in muscle myostatin, cytokines, and body composition were significantly greater in prediabetes than in RA, consistent with impaired muscle remodeling in RA. To maximize physiologic improvements with exercise training in RA, a better understanding is needed of skeletal muscle and physiologic responses to exercise training and their modulation by RA disease-specific features or pharmacologic agents or both. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT02528344 . Registered on August 19, 2015.