Powered lower limb prosthesis technologies hold the promise of providing greater ability and mobility to transfemoral amputees. Intent recognition systems for these devices may allow amputees to perform automatic, seamless transitions between locomotion modes. Prior studies in which pattern recognition algorithms have been trained to recognize subject-specific patterns within device-mounted sensor data have shown the feasibility of such systems. While effective, these strategies require substantial training regimens. To reduce this training burden, we developed and evaluated user-independent intent recognition systems. A novel mode-specific classification system was developed that allowed each locomotion transition to be statistically considered its own class. Various pattern recognition algorithms were trained with sensor data from a pool of eight lower limb amputees and performance was tested using data on a novel subject. For both user-dependent and user-independent classification, mode-specific classification reduced error ( ) on transitional steps by ∼ 50% without affecting steady-state classification. Incorporating sensor time history and level-ground walking data from the novel subject into the training data resulted in decreasing errors ( ) on steady-state classification by over 60% without affecting transitional error. These strategies were combined to demonstrate significant overall system improvements from baseline conditions presented in prior research.