Obstacles in the path of a running robot need to be avoided in order to avoid falling down. Currently, there are no control strategies that determine the appropriate step locations to obtain a successful overstep of an obstacle. The objective of this simulation study is to maximize the gap size by determining the step locations. The step strategy is tested on the SLIP-model and a model containing leg damping and push-off. With the means of an optimization, it is found that the optimal step strategy consists out of 3 phases: an adaptive phase from running cycle to the optimal state for the beginning of the leap, the beginning and end of the leap and another adaptive phase to end in a desired end state. For the SLIP-model is found that the maximum gap size is almost independent of the initial velocity of the model and mostly depends of the system’s energy. In order to maximize the gap size, the first and second step location have to be coincident. Furthermore, the damping and push-off proved to be an important factor for the step locations and the obtained gap size, as the first and second step locations do not coincident anymore and the gap size is reduced significantly.