Examination of a contemporary skeletal collection revealed a rheumatoid subgroup with parameters mirroring those of contemporary clinical populations. This rheumatoid population was also indistinguishable from contemporary (live) clinical populations, on the basis of the actual distribution of radiologically detectable erosions, thus validating its representativeness. Gross examination of these defleshed skeletons allowed assessment of the significance of erosive disease in rheumatoid arthritis. As anticipated, the frequency of visibly detectable erosions exceeded that detectable radiologically. The frequency of visibly detectable erosions, however, was indistinguishable from the frequency of synovitis in clinical populations. The excellent correlation of the gross and clinical distribution of disease suggests that some degree of erosive disease is integral to all lesions of rheumatoid arthritis and that only the relative insensitivity of radiologic techniques precludes universal recognition of those erosions. This work suggests that any therapeutic intervention which settles for only partial synovitis control will not prevent the progression of erosive disease; and perhaps explains the difficulty of demonstrating alterations in erosion progression in studies of disease modifying agents.