Objective: To assess the test-retest reliability and discriminative validity of a 14-item manual wheelchair circuit adapted from previous research (AMWC). Design: Two AMWC trials per subject completed within 15 days. Setting: Two clinical research and 3 rehabilitation centers. Participants: Convenience sample of individuals with spinal cord injury (N=50) from centers in the United States (n=38) and the Netherlands (n=12). Mean age ± SD was 46±13 years, and mean injury duration ± SD was 12±11 years. Fifteen had cervical injuries, and 42 were men. Interventions: An existing 8-task manual wheelchair circuit was modified to remove the need for a wheelchair treadmill and expanded to 14 tasks to attenuate floor and ceiling effects: 5 original tasks - figure-of-8, .012-m doorstep crossing, .10-m platform, 15-m sprint, and making a level transfer; 3 modified tasks - 3% and 6% ramp, and 3-minute overground wheeling; and 6 new tasks - .04-m doorstep crossing, propelling over artificial grass, opening/closing a door, 3% side slope, holding a wheelie for 10 seconds, and propelling in a wheelie. Main Outcome Measures: Reliability of the primary outcomes, sum ability score (sum of all tasks; 0-14 [no.]) and sum performance time (figure-of-8 + sprint + grass; 0-360 [s]), was determined by intraclass correlation coefficients (ICCs) for the whole sample and paraplegia (PP) and tetraplegia (TP) subsets. Independent t tests compared PP and TP trial 1 sum ability score and sum performance time. Results: Sum ability and sum performance time ICCs exceeded .90 for the full sample and the PP/TP subsets. Sum ability was higher for PP than TP (PP, 12.9±1.2; TP, 9.8±2.8; P<.00), and sum performance times were lower for PP than TP (20.0±4.0s vs 32.0±1.97s, P<.00). Conclusions: AMWC primary outcomes, sum ability score and sum performance time, are reliable and discriminate between TP and PP. © 2011 American Congress of Rehabilitation Medicine.