The main objective of the work is to address the estimation of the motor competence of girls and boys from the spontaneity of the motor game. 112 individuals of mixed groups were selected (M = 14.2, SD = 2.2) for the practice of two motor games. An observational methodology was used through a nomothetic, following and multidimensional design, with the assistance of two expert observers for the registration. The registration tool consisted of 19 categories, grouped into 3 macrocriteria. All the decisions of the protagonists were classified according to their gender, game role, quality of motor behavior and cooperative or antagonistic sense of the decision observed. Chi-square and multivariate analysis were applied (Chaid Model), being variables Cartesian tested in their double function (VD / VI). Among the most relevant results, revealing a differentiated participation according to the gender of the protagonists, girls registered similar values in effective behaviors in both games (J1 = 186, J5 = 183) unlike the boys (J1 = 240, J5 = 307). The predictive capacity of motor decisions when playing throught the valence (p <.001) and role (p <.001) among others, were the most determinant variables. Evaluating motor competence through motor games is viable without altering the validity of the curriculum and the spontaneity of the contextualized decision.