Empirical evidence provides support for the inclusion of yoga as part of eating disorder prevention efforts through its positive impact on positive embodiment and experience of positive core affect. However, there is a need to identify the specific instructional strategies that will more consistently support positive embodiment and positive affect. We examined the effect of teaching a single yoga class using mindfulness-based instruction compared to appearance-based and neutral instruction alternatives on embodiment (i.e., state body surveillance, state body appreciation, pleasure during yoga) and changes in affect from before to after class. Female participants ( = 62;  = 23.89,  = 6.86) were randomly assigned to a yoga class that emphasized: being mindfully present in one's body, changing one's appearance, or just getting into yoga poses. ANOVAs revealed significantly higher body surveillance (η =.10) and lower forecasted pleasure (η =.21) in the appearance class compared to the other two classes. Participants in the mindfulness class experienced greater improvement in affect (η =.08) from before to after class and higher remembered pleasure during the yoga class (η =.19) compared to those in the appearance class. Emphasizing changes to appearance in yoga instruction may place participants at risk for less positive affect and less positive experiences of embodiment compared to mindfulness-based or even neutral yoga instruction.