Objective To determine the optimal cutoff scores for the Fugl-Meyer Assessment of the Upper Extremity (FMA-UE) with regard to predicting no, poor, limited, notable, or full upper-limb capacity according to frequently used cutoff points for the Action Research Arm Test (ARAT) at 6 months poststroke. Design Prospective. Setting Rehabilitation center. Participants Patients (N=460) with a first-ever ischemic stroke at 6 months poststroke. Interventions Not applicable. Main Outcome Measures Based on the ARAT classification of poor to full upper-limb capacity, receiver operating characteristic curves were used to calculate the area under the curve, optimal cutoff points for the FMA-UE were determined, and a weighted kappa was used to assess the agreement. Results FMA-UE scores of 0 through 22 represent no upper-limb capacity (ARAT 0-10); scores of 23 through 31 represent poor capacity (ARAT 11-21); scores of 32 through 47 represent limited capacity (ARAT 22-42); scores of 48 through 52 represent notable capacity (ARAT 43-54); and scores of 53 through 66 represent full upper-limb capacity (ARAT 55-57). Overall, areas under the curve ranged from.916 (95% confidence interval [CI],.890-.943) to.988 (95% CI,.978-.998; P<.001). Conclusions There is considerable overlap in the area under the curve between the ARAT and FMA-UE. FMA-UE scores >31 points correspond to no to poor arm-hand capacity (ie, ≤21 points) on the ARAT, whereas FMA-UE scores >31 correspond to limited to full arm-hand capacity (ie, >22 points) on the ARAT.