The study of stroke and its treatment in human subjects requires accurate measurement of behavioral status. Arm motor deficits are among the most common sequelae after stroke. The Action Research Arm Test (ARAT) is a reliable, valid measure of arm motor status after stroke. This test has established value for characterizing clinical state and for measuring spontaneous and therapy-induced recovery; however, sufficient details have not been previously published to allow for performance of this scale in a standardized manner over time and across sites. Such an approach to ARAT scoring would likely reduce variance between investigators and sites. This report therefore includes a manual that provides a highly detailed and standardized approach for assigning ARAT scores. Intrarater reliability and interrater reliability, as well as validity, with this approach were measured and are excellent. The ARAT, when performed in a standardized manner, is a useful tool for assessment of arm motor deficits after stroke.