The physical self-concept is considered a significant predictor of physical activity and persistence in childhood and youth. Based on Shavelson, Stanton, and Hubner's (1976) hierarchical structure of self-concept, several approaches have been developed to measure physical self-perception. A multi-dimensional fitness-based approach by Marsh (1990) shows a valid method for measuring physical self-concept in all age groups from 8 years old to adulthood. Furthermore, following standard childhood fitness test dimensions, instruments have been developed for middle to late childhood (Dreiskämper, Tietjens, Honemann, Naul, & Freund, 2015a) and early childhood (Tietjens et al., 2018). However, based on Harter's (1980) approach of a more global and general skills-summarizing self-perception profile, other studies – preliminary coming from a motor development perspective – focused on measuring children's self-perception based on fundamental movement skills such as object control, locomotion, and stability. In this line, Estevan and Barnett (2018) proposed an adaption of Fox and Corbin's (1989) model. However, several research questions have remained unanswered because these different research directions have not been integrated yet: 1) How is the physical self-concept in childhood constructed, and how can it be measured? 2) How does the physical self-concept develop across childhood (and is this consistent with the assumptions about the self-concept as defined by Shavelson et al., 1976)? And 3) What is the role of physical self-concept in behavior (i.e., PA) and its antecedents in childhood (and how)? The purpose of this discussion paper is to address these three more or less open questions from a self-concept research perspective. To this end, the models of Shavelson et al. (1976) and Marsh (1990) will be drawn upon to integrate the recent research developments into a new classification of the physical self-concept and its development throughout childhood.