OBJECTIVES: To determine preliminary efficacy and to identify baseline characteristics predicting who would benefit most from fast walking training plus a step activity monitoring program (FAST+SAM) compared with fast walking training (FAST) alone in persons with chronic stroke. DESIGN: Randomized controlled trial with blinded assessors. SETTING: Outpatient clinical research laboratory. PARTICIPANTS: Individuals (N=37) >6 months poststroke. INTERVENTIONS: Subjects were assigned to either FAST, which was walking training at their fastest possible speed on the treadmill (30min) and overground 3 times per week for 12 weeks, or FAST+SAM. The step activity monitoring program consisted of daily step monitoring with an activity monitor, goal setting, and identification of barriers to activity and strategies to overcome barriers. MAIN OUTCOME MEASURES: Daily step activity metrics (steps/day [SPD], time walking per day), walking speed, and 6-minute walk test (6MWT) distance. RESULTS: There was a significant effect of time for both groups, with all outcomes improving from pre- to posttraining (all P values <.05). The FAST+SAM was superior to FAST for 6MWT (P=.018), with a larger increase in the FAST+SAM group. The interventions had differential effectiveness based on baseline step activity. Sequential moderated regression models demonstrated that for subjects with baseline levels of step activity and 6MWT distances that were below the mean, the FAST+SAM intervention was more effective than FAST (1715±1584 vs 254±933 SPD; P<.05 for overall model and ΔR(2) for SPD and 6MWT). CONCLUSIONS: The addition of a step activity monitoring program to a fast walking training intervention may be most effective in persons with chronic stroke who have initial low levels of walking endurance and activity. Regardless of baseline performance, the FAST+SAM intervention was more effective for improving walking endurance.