Aim: To determine if treadmill training embedded in self-management education commencing during stroke inpatient rehabilitation results in more physical activity than usual gait training. Method: A prospective, parallel-group, randomised trial with concealed allocation, blinded measurement and intention-to-treat analysis involving 119 stroke survivors undergoing rehabilitation who were able to walk independently was undertaken. The experimental group undertook treadmill training (40-60% heart rate reserve) and self-management education for 30 minutes, three times a week for 8 weeks and the control group undertook the same amount of usual gait training. Outcomes were measured at baseline (Week 0), on completion of the intervention (Week 8) and beyond the intervention (Week 26). The primary outcome was physical activity measured as steps/day using an activity monitor. Secondary outcomes were walking ability, cardiorespiratory fitness, cardiovascular risk, depression, self-efficacy, perception of physical activity, participation, and quality of life. Results: After 8 weeks, the experimental group took 1436 more steps/day (95% CI 229 to 2643) than the control group. By 6 months, they took 871 more steps/day (95% CI -385 to 2129) than the control group. There was no difference between groups in any other outcome. Conclusions: In individuals undergoing rehabilitation after stroke, 8 weeks of treadmill training embedded in self-management resulted in more physical activity than usual gait training and this was largely maintained at 6 months, despite little effect on walking or cardiorespiratory fitness, suggesting the self-management was responsible. Trial Registration: ACTRN12613000744752. Data Access: Data are available from the corresponding author.