Body-weight-supported locomotor training has been studied for three decades, yet its clinical outcomes remain inconsistent. The standard interpretation of this inconsistency is that the right dosage, timing, or patient subgroup has not been found. We propose a different reading: the inconsistency may reflect a conceptual underspecification in how loading is understood. Two interpretations of body-weight support coexist in the literature—loading as task difficulty (unloading makes the task easier) and loading as control strategy (unloading changes which neural circuits can participate in the solution). These two interpretations generate different predictions, imply different training protocols, and have rarely been distinguished empirically. We review evidence from spinal reflex physiology, body-weight-supported treadmill training trials, and developmental motor control, and argue that the strategy interpretation is better supported but has not been directly tested against the difficulty interpretation. Three testable predictions are offered that could distinguish the two. The framework has direct implications for the sequencing of load in rehabilitation and for the interpretation of load-dependent training failure.