To establish: (1) inter-rater and test-retest reliability of standardized guidelines for the Fugl-Meyer upper limb section, Action Research Arm Test and Box and Block Test in patients with paresis secondary to stroke, multiple sclerosis or traumatic brain injury and (2) correlation between these arm motor scales and more general measures of impairment and activity limitation. Multicentre cohort study. Three European referral centres for neurorehabilitation. Thirty-seven stroke, 14 multiple sclerosis and five traumatic brain injury patients. Scores of the Fugl-Meyer Test (arm section), Action Research Arm Test, and Box and Block Test derived from video information. All three motor tests showed very high inter-rater and test-retest reliability (ICC and rho for main variables > 0.95). Correlation between the motor scales was very high (rho > 0.92). Motor scales correlated moderately highly with the Hemispheric Stroke Scale, a measure of impairment (rho = 0.660-0.689), but not with the Modified Barthel Index, a measure of the ability to cope with basic activities of daily living (rho = 0.044-0.086). The standardized guidelines assured comparability of test administration and scoring across clinical facilities. The arm motor scales provided information that was not identical to information from the Hemispheric Stroke Scale or the Modified Barthel Index.