Context:Increased mechanical loading can promote the preferential differentiation of bone marrow mesenchymal stem cells to osteoblastogenesis, but it is not known whether long-term bone strength-enhancing exercise in humans can reduce marrow adiposity.Objective:Our objective was to examine whether bone marrow density (MaD), as an estimate of marrow adiposity 1) differs between young female athletes with contrasting loading histories and bone strengths and 2) is an independent predictor of bone strength at the weight-bearing tibia.Design:Mid-tibial MaD, cortical area (CoA), total area, medullary area, strength strain index (SSI), and cortical volumetric bone mineral density (vBMD) (total, endocortical, midcortical, and pericortical) was assessed using peripheral quantitative computed tomography in 179 female athletes involved in both impact and nonimpact loading sports and 41 controls aged 17-40 years.Results:As we have previously reported CoA, total area, and SSI were 16% to 24% greater in the impact group compared with the controls (all P < .001) and 12% to 18% greater than in the nonimpact group (all P < .001). The impact group also had 0.5% higher MaD than the nonimpact and control groups (both P < .05). Regression analysis further showed that midtibial MaD was significantly associated with SSI, CoA, endocortical vBMD, and pericortical vBMD (P < .05) in all women combined, after adjusting for age, bone length, loading groups, medullary area, muscle cross-sectional area, and percent fat.Conclusion:In young female athletes, tibial bone MaD was associated with loading history and was an independent predictor of tibial bone strength. These findings suggest that an exercise-induced increase in bone strength may be mediated via reduced bone marrow adiposity and consequently increased osteoblastogenesis.