Safe knee function under all real-world conditions is a crucial factor in the prescription of specific prosthetic knee mechanisms. Although many amputees have reported the subjective belief that the electronic C-Leg combines increased function while walking with increased safety, till date there has been little objective evidence to support this claim. This study was designed to identify biomechanical differences that would permit objective evaluation of the relative safety in critical situations of different prosthetic knee joint designs. In the gait laboratory, three experienced amputees wearing a safety harness were tested under conditions that simulated five common real-world situations: walking on even ground, abruptly stopping, abruptly sidestepping, inadvertently stepping onto an object, and tripping when the knee is extending during swing phase. Kinematics and kinetics of three knee joints—the 3C1 (Mauch SNS hydraulic system), 3R80 (rotary hydraulic system), and C-Leg (electronically controlled hydraulic system)—were measured using accepted gait analysis technology (Kistler force plates, VICON system). This study protocol proved to be suitable for defining the potential safety of tested knee joints. The results from instrumented gait analysis were shown to provide an objective reason for knee stability or instability for each individual trial, including the reason for knee collapse. The amputee subjects confirmed that the gait disruption occurring under the tested conditions corresponded closely to critical everyday situations that may lead to fall. Tripping with 3R80 and stepping onto an object with 3C1 results in a significant risk of falling. Because of its biomechanical performance under high-risk conditions, C-Leg seems to be the most suitable design to prevent falls with the prosthesis.