The committee was charged with determining whether healing of erosions in rheumatoid arthritis (RA) occurs. Two exercises were performed: The first asked the committee members, as a panel of experts, to express agreement or disagreement with the presence of improvement and features of bone reaction to injury in images submitted by members as examples of healing. The second presented panel members with 28 pairs of serial images, 14 chosen to illustrate progression and 14 chosen to illustrate repair. Agreement was tested on 8 items: global judgment on which image in the pair was better, relative size of the erosion in the 2 images, judgment on which image was first, presence and extent of sclerosis, cortication, filling-in, remodeling, and reconstituting normal structure. Our results showed good agreement, among the 15 respondents, on global assessment of which image was better and which image showed the smaller erosion. Correct assignment of sequence was only slightly better than expected by chance (in 65% of the cases). Agreement was poor regarding the presence of morphologic features of bone repair. A majority of a panel of experts agreed on which 2nd images in a set of paired, serial images represented improvement and which showed progression based on global assessment of which was better and on size of erosion. Features of bone repair were not distinctive and did not enable the panel to deduce the correct sequence of the serial images. This study provides evidence that repair of bone damage in RA does occur, resulting in some degree of improvement, which was recognized by a majority of a panel of experts.