Objective: To continue and expand determination of the re liability, validity, and sensitivity to change of the Arm Motor Ability Test (AMAT), an instrument for assessing deficits in activities of daily living (ADL). Design: The AMAT was administered twice to patients, with an interest interval of either 1 or 2 weeks, by one of two examiners assigned to patients in counterbalanced order. Patients' interest intervals and scores on the arm portion of the Motricity Index was unknown to the raters. Setting: A referral inpatient neurological rehabilitation center. Patients: Thirty three subacute stroke inpatients with moderate to mild upper extremity motor deficit: median Motricity-Index-Arm score = 89, median chronicity = 43d, median age = 66 yr; 12 were women. Main Outcome Measure and Results: The AMAT was developed in 1987, and interrater reliabilities at that time were found to range from .95 to .99. The present values for interrater reliability (2 scales) from videotaped test performance were: kappas =.68 to .77, Spearman correlations = .97 to .99. For performance time, interscorer reliability from videotaped test performance was .99. Homogeneities for the three AMAT measures for the total sample (Cronbach's alpha and split-half reliability) were .93 to .99. The test-retest reliabilities for the total sample were .93 to .99. The correlations to the Motricity-Index-Arm score were .45 to .61. The AMAT detected the difference in change occurring as a result of the passage of 1 versus 2 weeks in these subacute inpatients, presumably as a result of intensive therapy and/or spontaneous recovery, confirming the results of an earlier intervention study. Conclusion: The AMAT is an instrument with high interrater reliability, internal consistency, and sensitivity to change, as well as having satisfactory concurrent validity.