We’re interested in the use of active tails for steady-state locomotion of legged systems. Building on the work of [Libby12] and [Johnson12] who thoroughly analysed active tail-use for body-pitch control during flight-phase, we focus on the stance phase through analysis of mathematical models, numerical optimisation as well as hardware testing on the cat-inspired robot Cheetah-Cub[Sproewitz13]. Starting with a SLIP-model [Blickhan89] augmented with an active flywheel, we find the main advantage of the additional control-input is in decoupling body-pitch stabilisation from the task of injecting energy into the system: all leg-actuators can thus be recruited for performing positive work on the body, while the flywheel maintains trunk stability. We hypothesise that this simplification of motor-control is also a key advantage when using a tail. However, in a more realistic model with a full tail, these control problems are coupled. We establish criteria for designing a tail that effectively decouples the two control problems and analyse their implications both analytically and through numerical optimisation. A long, light tail optimises these criteria and results from simulations and hardware tests match this prediction. We find that for small animals the main constraint is the range of displacement of the tail, similar to [Johnson12]. As we scale upwards the constraint becomes actuation power: the muscle-content of the tail necessary to keep up should scale at a power of 4/3s with body mass, which conflicts with the light-and-long design. Hence the effectiveness of a tail to decouple control becomes limited as we scale upwards. We match this with selected biological data [Alexander75] as well as abstract cases, and find that numerical optimisations generally match with the predictions.