The responsiveness of the Action Research Arm (ARA) test and the upper extremity motor section of the Fugl-Meyer Assessment (FMA) scale were compared in a cohort of 22 chronic stroke patients undergoing intensive forced use treatment aimed at improvement of upper extremity function. The cohort consisted of 13 men and 9 women, median age 58.5 years, median time since stroke 3.6 years. Responsiveness was defined as the sensitivity of an instrument to real change. Two baseline measurements were performed with a 2-week interval before the intervention, and a follow-up measurement after 2 weeks of intensive forced use treatment. The limits of agreement, according to the Bland-Altman method, were computed as a measure of the test-retest reliability. Two different measures of responsiveness were compared: (i) the number of patients who improved more than the upper limit of agreement during the intervention; (ii) the responsiveness ratio. The limits of agreement, designating the interval comprising 95% of the differences between two measurements in a stable individual, were -5.7 to 6.2 and -5.0 to 6.6 for the ARA test and the FMA scale, respectively. The possible sum scores range from 0 to 57 (ARA) and from 0 to 66 (FMA). The number of patients who improved more than the upper limit were 12 (54.5%) and 2 (9.1%); and the responsiveness ratios were 2.03 and 0.41 for the ARA test and the FMA scale, respectively. These results strongly suggest that the ARA test is more responsive to improvement in upper extremity function than the FMA scale in chronic stroke patients undergoing forced use treatment.