This study demonstrated that implementation of the Proteus effect via manipulation of avatar age in VR is effective among elderly individuals in the context of exercise. One hundred and four elderly adults aged 60 years and older who did not engage in vigorous physical activities participated in this experiment with a 2 (avatar age: young vs. older) × 2 (sex: male vs. female) design. The results showed that the embodiment of younger avatars (age approximately 20 years) in VR leads to greater perceived exercise exertion regardless of sex after controlling for age and emotion. Older adults with young avatars perceived a greater contribution of efforts to exercise. This study also found that among those who did not engage in vigorous exercise, female older adults who embodied young avatars reported greater self-efficacy for future exercise and greater physical activity during the exercise phase than those who embodied older avatars. This study suggests that females are more likely to be motivated to continue exercising through young avatar embodiment. In contrast, female elderly who embodied old avatars reported significant fewer physical activity than male elderly who embodied old avatars. This indicated that the Proteus effect had stronger effects among females than among older males. Although we found the Proteus effect through VR avatar manipulation, the effect was temporary and limited to the experimental phase. This study is the first to examine the Proteus effect among elderly individuals in the context of exercise. It also contributes to the literature by indicating that avatar age manipulation is an effective means of promoting exercise among elderly individuals and helping them achieve exercise outcomes. This study further demonstrates that female elderly individuals respond to young avatars differently than male elderly individuals, with female elderly individuals showing more positive effects of young avatar embodiment than males. Implications and theoretical contributions are discussed.