This study investigated the effect of treadmill walking training with and without random speed changes in persons in the chronic stage of stroke. It was hypothesized that random speed change group will show less attention and balance demands for speed change in level walking.

This study includes chronic stroke patients who will be randomly assigned to blocked or random walking speed change group to receive 30 min per session, 3 sessions per week for 2 weeks' walking training on a treadmill. The outcome measures include gait characteristics under 3 tasks x 3 speed conditions x 2 attention conditions on an obstacle-free walk way. The three tasks are normal walking, narrow based walking and obstacle crossing. The three speed conditions are constant (preferred) speed, speeding up (preferred to maximum speed), and slowing down (maximum to preferred speed). The two attention conditions are none and serial subtraction of 3 during walking. Clinical sensorimotor function assessment, including hand grip strength, plantar sensitivity, stroke rehabilitation assessment of movement and Fugl-Meyer lower extremity motor scale will be conducted for subject characterization.

Inclusion Criteria:

• Patients diagnosed with first stroke for at least 6 months
• Able to walk for at least 20 minutes independently without exhaustion
• Residual gait deficits
• Resting blood pressure is lower than 150/90 mmHg

Exclusion Criteria:

• Unable to follow experimental instructions during the process
• Any pain, inflammation, or other neuromuscular or musculoskeletal conditions in the lower limbs that affect walking ability
• Unstable health problems