The validation of measuring instruments in the field of health is a requirement before they can be used safely and reliably. The action research arm test (ARAT) tool is an instrument validated in numerous countries and languages and for different populations, and its use is widespread. The objective of this research was to determine the psychometric properties of ARAT for a sample composed of post-stroke patients. To achieve this, a psychometric analysis was performed, where internal consistency tests were carried out using Cronbach’s alpha, correlations between items and item-total and half-level tests to verify their reliability. Regarding validity, criteria validity tests were performed, taking the motor function dimension of the Fugl–Meyer scale as gold standard, and convergent validity tests were performed by correlation with the FIM–FAM, ECVI-38 and Lawton and Brody scales. The results showed very good internal consistency as well as good criterion and convergent validity. In conclusion, the ARAT can be considered a valid and reliable instrument for the evaluation of upper limb function in post-stroke patients.