Objectives. The main objectives of this experimental case series were to evaluate the feasibility of a modified dance-based exercise program with low ground impacts in persons with rheumatoid arthritis (RA) functional class III and to describe its effects on locomotor ability. The relationship between 3 loco-motor tests and their responsiveness also were addressed. Methods. Ten female subjects participated in an 8-week exercise program. Locomotor ability was measured before and after the program using the 50-foot test of walking time, the 6-minute test of walking distance, and the locomotion biomechani-cal analysis. Results. All subjects showed a high compliance (92.5% presence at sessions) over the 8 weeks of exercise without any aggravation in disease status. They were able to train efficiently at moderate intensity up to 25 minutes. Significant improvements were found in locomotor ability, with a higher responsiveness measured by the locomotion biome-chanical analysis, followed by the 6-minute gait test and the 50-foot gait test. Inconsistent relationships between tests suggested that different locomotor abilities are required during tests. Conclusion. These results support the feasibility of a modified dance-based exercise program for persons with severe RA. With high levels of responsiveness , the detailed biomechanical analysis and the 6-minute gait test are recommended for the assessment of locomotor ability.