The aims of this study were to compare the load of three tasks designed to train features of soccer: a transition game, a small-sided game with a change of playing area, and a large-sided game. Twenty young elite players performed these tasks. Variables measured were total distance covered (DC), distance covered at 14.0-17.9 km·h-1 (DC 14.0-17.9 km·h-1), distance covered at 18.0-21 km·h-1 (DC 18-21 km·h-1), distance covered at >21 km·h-1 (DC >21 km·h-1), peak speed, accelerations and decelerations >1.0 and >2.5 m·s-2, player load, and rate of perceived exertion. Transition games produced greater DC 18-21 km·h-1, DC >21 km·h-1, peak speed and Acc >2.5 m·s-2 than the other drills (p<0.01) and higher DC 14-17.9 km·h<sup>-1</sup> (p<0.01), Dec >1 m·s<sup>-2</sup> (p<0.05) and Dec >2.5 m·s<sup>-2</sup> (p<0.01) than the large-sided game. Both sided games produced more DC (p<0.01), Acc >1 m·s<sup>-2</sup> (p<0.01) and player load (p<0.01 and p<0.05, respectively) than the transition game. During the small-sided game, significantly higher DC 14-17.9 km·h<sup>-1</sup>, DC 18-21 km·h<sup>-1</sup> and DC >21 km·h<sup>-1</sup> were recorded in comparison with the large-sided game (p<0.01). The studied parameters showed lower variation in the transition game. Coaches could use transition games to train high speed running in counter-attack contexts.