A simple measure of overall walking effort would be valuable to patients and clinicians to select suitable treatment interventions and to monitor progress. In this paper, the reproducibility and responsiveness of seven potentially useful clinical measures of walking effort are presented. These walking effort outcomes were derived from a compass gait model and space curve displacement, acceleration and differential geometry theory. The walking effort outcomes were primarily calculated from the motion of a point on or in the rigid body pelvis as a patient walked cyclically. These motion data were collected from eight healthy volunteers who each walked on a treadmill for 4 min, at four different speeds, repeated twice. Four of the seven walking effort outcomes clearly had better measurement properties. The path length ratio, acceleration ratio, Frenet–Serret torsion and Frenet–Serret energy had excellent reproducibility (ICC>0.8) and responded to a small change in walking speed (< 0.03 m/s) compared to two versions of the biomechanical efficiency quotient and the Frenet–Serret curvature. The measurement properties of most outcomes were not consistently improved using a point in versus on the pelvis. This study presents four biomechanical walking effort outcomes that have good theoretical underpinnings, excellent reproducibility and responsiveness, are simple and easy to administer with relatively inexpensive equipment, and can be used in real world environments. However, future work must investigate the minimal clinically important change of these outcome measures before they can be used in clinical practice.