As we age, reliance on the ankle musculature for push‐off during walking reduces and increased reliance on the hip musculature is observed. It is unclear how joint pathology like osteoarthritis may affect this distal‐to‐proximal redistribution of propulsion. Here, we revisited a proof‐of‐concept study to study the effect of split‐belt treadmill training, designed to reduce step length asymmetry, on forward propulsion during walking. Eleven women with hip osteoarthritis and five age‐matched control participants walked on an instrumented split‐belt treadmill at their preferred speed (hip osteoarthritis: 0.73 ± 0.11 m/s; controls: 0.59 ± 0.26 m/s). Women with hip osteoarthritis had less ankle power and propulsive force than controls, and greater hip contributions to forward propulsion on their involved limb. Following split‐belt treadmill training, propulsive force increased on the involved limb. Five of 11 participants experienced a change in redistribution ratio that was greater than the minimal clinically meaningful difference. These “responders” had greater variability in pre‐training redistribution ratio compared to non‐responders. Women with hip osteoarthritis had poorer propulsive gait mechanics than controls yet split‐belt treadmill training improved propulsive force. Redistribution ratio also changed in participants with high baseline variability. Our results suggest that split‐belt treadmill training may be beneficial to people with hip osteoarthritis who have high variability in walking parameters. Further, the age‐related shift to increased hip contributions to propulsion across populations of older adults may be due to increased variability during walking.