We have re-examined the value of hand radiographs in the assessment of joint damage in rheumatoid arthritis in a series of related studies. Our objectives were: to assess the reproducibility of scoring hand radiographs and the relationships between different types of changes; to analyse the correlations between radiographic changes and hand function; and to assess the inter-relation between hand changes and overall joint damage. We confirmed the reproducibility of radiological assessments and showed that two observers can reproducibly score hand radiographs separately for joint space loss and erosions; both of these correlated with the Larsen score derived from standard radiographs. However, hand radiographic changes were not closely related to hand function; nor did they predict damage to large joints. Finally, examination of changes in 13 joint groups showed that there is a weak relationship between damage and disease duration, and the rate of damage is greatest in the initial years. We conclude that the status of hand radiography as a measure of overall outcome in rheumatoid arthritis remains an open question.