Twenty-four patients with stroke were studied (a) to determine the interrater reliability of a clinical measurement of shoulder subluxation, (b) to confirm the interrater reliability of the Ritchie Articular Index (Bohannon & LeFort, 1986) for measuring shoulder pain, (c) to establish the relationship between the Ritchie index scores and shoulder lateral rotation range of motion measured at the point of pain (SROMP), and (d) to determine the relationship between shoulder subluxation and shoulder pain. The agreement between the two examiners' (the authors) measurements of subluxation was "almost perfect" (Landis & Koch, 1977, p. 165). The agreement between the two examiners' Ritchie index measurements was "substantial" (Landis & Koch, 1977, p. 165). The Ritchie index and SROMP measurements correlated significantly. Neither the Ritchie index nor the SROMP measurements correlated significantly with subluxation. Although the measurements used in this study were reliable, they did not support the association of shoulder subluxation with shoulder pain in stroke patients. Clinicians wishing to reduce shoulder pain in stroke patients should direct their treatment accordingly.