Recently developed motorized lower limb prostheses which provide joint power at the knee and/or ankle have been shown to assist users in performing a variety of ambulation tasks. Control of these devices, specifically creating seamless, natural, and accurate mode transitions, is challenging. The control system also needs to allow users the flexibility to perform these tasks with a wide range of variability. The goal of this study was to develop and test a flexible control system for a powered knee-Ankle prosthesis. Specifically, a pattern recognition system was trained to recognize the transitions between six modes (standing, level-ground walking, ramp ascent, ramp descent, stair ascent, and stair descent) at various discrete time points during the gait cycle. Six transfemoral users were successful performing a variety of ambulation tasks including climbing a 3-and 4-step staircases, performing walking and standing transitions to and from the staircases, approaching the stairs and ramps at various angles, and ambulating at various speeds. Offline classification results demonstrate that system error rates 1.6% [1.0%], mean [standard deviation], at heel contact and 0.5% [0.2%] at toe off. The results of this study are encouraging and provide another step towards a clinically viable lower limb intent recognition system for a powered knee-Ankle prosthesis.