To determine the histological indication for arthroscopic synovectomy in rheumatoid knees, 23 patients underwent lavage and biopsy by needle arthroscopy. Eighty-one patients were treated with arthroscopic knee synovectomy after needle arthroscopy, and 51 of these patients underwent only arthroscopic synovectomy. Thirty patients who showed no improvement following arthroscopic synovectomy underwent open surgical synovectomy. Intra-articular lavage of knee joints by needle arthroscopy did not result in any clinical improvements. In the patients who underwent only arthroscopic synovectomy, two groups (only macrophages, no macrophages or B cells) showed clinical improvement (symptoms of the knee, ESR, RF). Patients in whom open surgical synovectomy was performed because of the failure of arthroscopic synovectomy showed clinical improvement. However, there was a significant loss of movement in the knee joint and more severe radiological deterioration in patients who underwent open surgical synovectomy than in those who underwent only arthroscopic synovectomy. Arthroscopic synovectomy should be used for rheumatoid patients with synovial tissue containing only macrophages or none of these two cells.