Rubble-strewn corridors, stairs and steep natural terrain all present a challenge for wheels and tracks. Legs are a solution in these cases because foot placement allows the traversal of discontinuous terrain. Legged robots, however, currently lack the performance needed for practical applications. This work seeks to address an aspect of the problem, foot placement while running. A novel hopping height controller for a spring-loaded legged robot is presented. It is simple and performs well enough to allow control of the ballistic trajectory of hops and therefore foot placement. Additionally, it can adapt to different ground properties using the result from previous hops to update control gains. A control strategy of extending the leg at a fixed rate during the stance phase and modulating the rate of extension on each hop was used to control the hopping height. The extension rate was then determined by a feed-forward + proportional control loop. This performed sufficiently well allowing the ballistic trajectory of hops to be controlled. In simulation, the spring-loaded inverted pendulum (SLIP) model was extended to include actuation and losses due to friction. The control strategy was developed using this model then, in a planar simulation, the controller was run to perform foot placement while running over a series of platforms which vary in their horizontal and vertical spacing. To experimentally validate and further develop the control strategy, a one-legged hopping robot, constrained to move vertically, was used. The leg had 2 links, hydraulically actuated hip and knee joints and a spring-loaded foot. Results showed that the controller developed could be used to perform hops of randomly varying size on grounds with different properties and while running on a treadmill at different speeds. As an aside, the dynamics of hydraulic actuators presented a problem for foot repositioning during flight using a simple PID controller. This was solved through the novel implementation, in hydraulics, of a `zero-vibration' (ZV) filter in a closed-loop. Simulation and experimental results demonstrating this are presented.