Changes in muscle-fat-composition affect physical performance and muscular function, like strength and power. The purpose of the present study was to investigate whether changes in soft tissue composition of the thigh and changes in muscle size and composition resulting from physical training were detectable with Dixon magnetic resonance imaging (MRI). A young and healthy subject population (n = 21, 29 ± 5 years) was split into a strength training (G_t, 11 subjects) and a control group (G_c, 10 subjects). The physical training intervention lasted over 13 weeks. Before and after this intervention a muscle performance exam and an MRI exam were conducted on all subjects. To evaluate muscle performance and the training effect, the jump height was measured using a mechanograph. Fascia, pure muscle and subcutaneous fat areas and proton density water fraction (PDWF) and proton density fat fraction (PDFF) of the left thigh were measured with a 6-point Dixon prototype MRI sequence. Muscle area changed by +7.1 ± 3.3% (p < 0.05) and +2.5 ± 5.6% (p > 0.05), and PDFF by -16.3 ± 10.4% (p < 0.05) and +5.4 ± 6.9% (p > 0.05) in G_t and G_c, respectively. Cross-sectional and longitudinal correlation coefficients R between PDFF and muscle performance were moderate (R = -0.43 and R = -0.51, respectively). The correlation was also moderate for muscle performance and a combined muscle fat per area ratio (R = -0.40 and R = -0.55, respectively). Dixon MRI is capable to measure training-related changes in muscle area and muscular fat. Both parameters correlate to muscle function. Muscle area per se does not always mirror functional parameters. Due to the complex interaction of muscle volume, muscle structure, and inter- and intramuscular coordination during muscle performance, multivariate muscle parameter models should be investigated in the future. Future studies will have to show if structural parameters mirror and explain functional muscle data both in the context of physical training and pathologies like sarcopenia.