The ecological validity of three representative games of the invasion games tactical problems (keeping possession, attacking the goal and scoring) for assessing the game performance of 21 U-10 youth soccer players was analyzed. Data were analyzed according to the year of competition of the players in U-10 categories (First or Second) and the level of expertise. Second year players game performance was significantly higher in decisions for keeping in the game that represented this tactical problem (U = 33, p =.051, r = .44), and in passing decisions for keeping in the attacking game (U = 33, p = .044, r = .42). The level of expertise correlated significantly with the game performance in getting-free decisions and executions (rho = .573, p = .007; rho = .620, p = .003) for keeping in the keeping game, and also in the getting-free executions for attacking in the scoring game (rho = .480, p = .028). Less skilled players showed significantly higher values in the "spectator player" behavior in the scoring game (rho = -.521, p = .015). The findings are discussed in relation to the ecological validity of the games for learning and talent assessments.