Objectives <p>To combine Dixon-based quantitative MRI with automated segmentation to quantify fat fraction (FF) and lean normalised muscle volume (LNV) in seven core muscles and to derive two composite scores, comparing highly active cyclists with a physically inactive group.</p> Methods <p>Cyclists (<i>n</i> = 84) and physically inactive volunteers (<i>n</i> = 85) underwent Dixon MRI of the lumbar spine and pelvis capturing images of the psoas major, iliacus, quadratus lumborum, erector spinae/multifidus, gluteus maximus, gluteus medius and gluteus minimus. Images were analysed using automated segmentation to determine the FF and LNV of each muscle and two composite core scores summarising mean fat fraction (FF_core) and total lean normalised volume (LNV_core) across muscles. Multiple linear regression models were conducted to compare groups while adjusting for age, sex and BMI.</p> Results <p>Cyclists showed lower FF and higher LNV than physically inactive participants across muscles and for both composite scores. Age was a significant predictor of FF in all muscles and in FF_core (all <i>p</i> &lt; 0.001). Activity group was a significant predictor of FF in all muscles except gluteus minimus and quadratus lumborum, while sex was significant for most muscles except psoas major. For LNV, activity group, sex and BMI were significant predictors of LNV_core and all muscles, whereas age showed weaker associations. Notably, FF_core showed balanced sensitivity to activity group, sex, age and BMI (ηp²≈0.20–0.25; <i>R</i>²≈0.64).</p> Conclusions <p>We identified muscle-specific differences in core composition between highly active cyclists and physically inactive adults, and FF_core provided the most broadly sensitive summary measure across activity and demographic/body-composition factors.</p>