A study of 100 patients with rheumatoid arthritis has shown good correlation amongst three rating scales when used for measuring severity of morning stiffness and severity of stiffness present at the time of interview. Duration of morning stiffness was found to correlate only moderately well with severity of morning stiffness and poorly with severity of stiffness present at the time of interview. Investigation of the patients' definitions of stiffness, with and without the aid of a list of descriptive words, indicated an inter-relationship of pain and limited movement in the majority. This linking of the two symptoms under the term 'stiffness' may explain why some objective methods of measurement do not appear to reflect the subjective stiffness of which patients complain.