The contribution of marrow adipose tissue (MAT) to skeletal fragility is poorly understood. Peroxisome proliferator-activated receptor (PPAR)γ agonists, associated with increased fractures in diabetic patients, increase MAT. Here, we asked whether exercise could limit the MAT accrual and increase bone formation in the setting of PPARγ agonist treatment. Eight-week-old female C57BL/6 mice were treated with 20-mg/kg · d rosiglitazone (Rosi) and compared with control (CTL) animals. Exercise groups ran 12 km/d when provided access to running wheels (CTL exercise [CTL-E], Rosi-E). After 6 weeks, femoral MAT (volume of lipid binder osmium) and tibial bone morphology were assessed by microcomputer tomography. Rosi was associated with 40% higher femur MAT volume compared with CTL (P < .0001). Exercise suppressed MAT volume by half in CTL-E mice compared with CTL (P < .01) and 19% in Rosi-E compared with Rosi (P < .0001). Rosi treatment increased fat markers perilipin and fatty acid synthase mRNA by 4-fold (P < .01). Exercise was associated with increased uncoupling protein 1 mRNA expression in both CTL-E and Rosi-E groups (P < .05), suggestive of increased brown fat. Rosi increased cortical porosity (P < .0001) but did not significantly impact trabecular or cortical bone quantity. Importantly, exercise induction of trabecular bone volume was not prevented by Rosi (CTL-E 21% > CTL, P < .05; Rosi-E 26% > Rosi, P < .01). In summary, despite the Rosi induction of MAT extending well into the femoral diaphysis, exercise was able to significantly suppress MAT volume and induce bone formation. Our results suggest that the impact of PPARγ agonists on bone and marrow health can be partially mitigated by exercise.