Radiologic assessment of progressive joint destruction in rheumatoid arthritis is generally considered to be the ultimate standard for evaluation of treatment. We compared alternative radiologic techniques by performing a randomized, controlled trial in which hand films of rheumatoid arthritis patients were read by several skilled observers. The number of joints evaluated (34 versus 18) was found to make relatively little difference, but the number of readers and their experience level was critical. Films should be read in pairs. Joint space narrowing and erosion scores were shown to contribute independent information. Use of recommended techniques can reduce the number of patients required and, thus, can reduce the cost of a clinical trial by more than half and can substantially increase the sensitivity and efficiency of a trial. Therefore, critical selection of the method of assessing study endpoint is of great importance.