The morphology of entheses (muscle/tendon attachment sites) on bones is routinely used in paleontological and bioarcheological studies to infer the physical activity patterns of ancient vertebrate species including hominins. However, such inferences have often been disputed due to limitations of the quantitative methods commonly employed and a lack of experimental evidence demonstrating direct effects of physical activity on entheseal morphology. Recently, we introduced a new and improved method of quantifying and analyzing entheseal morphology that involves repeatable three-dimensional measurements combined with multivariate statistics focused on associations among multiple entheses. Here, to assess the validity of our method for investigating variation in entheseal morphology related to physical activity patterns, we analyzed femora of growing turkeys that were experimentally exercised for 10 weeks on either an inclined or declined treadmill or served as controls (n= 15 specimens, 5/group). Our multivariate approach identified certain patterns involving three different entheses (associated with muscles gluteus primus, medial gastrocnemius, vastus medialis and adductor magnus) that clearly differentiated controls from runners. Importantly, these differences were not observable when comparing groups within each of the three entheseal structures separately. Body mass was not correlated with the resulting multivariate patterns. These results provide the first experimental evidence that variation in physical activity patterns has a direct influence on entheseal morphology. Moreover, our findings highlight the promise of our newly-developed quantitative methods for analyzing the morphology of entheses to reconstruct the behavior of extinct vertebrate species based on their skeletal remains.