We tested patients with a well defined connective tissue disease (CTD) against 3 different sets of criteria for mixed connective tissue disease (MCTD). Included were 200 patients with systemic lupus erythematosus (SLE), 80 with MCTD, 100 with rheumatoid arthritis (RA), 80 with scleroderma, 53 with dermato/polymyositis (DM/PM) and 80 with primary Sjögren's syndrome (SS). The 3 sets of criteria fared similarly in capturing nearly all MCTD patients. They also were similar in ruling out most of the other CTD except for 11 patients with SLE, 36 with scleroderma, 13 with DM/PM and 3 with SS who fulfilled the category of possible MCTD included in the set of criteria proposed by Sharp. Because the set of criteria we proposed includes only 5 clinical manifestations (edema of the hands, synovitis, myositis, Raynaud's phenomenon, and acrosclerosis) whereas the other 2 sets include 15 and 13, respectively, it would seem that the 5 included in our criteria are core manifestations of MCTD. Of the 80 patients with MCTD 32 had all 5, 38 had 4, and 10 had 3 of these manifestations. The sensitivity of 3 or more of these clinical criteria for MCTD was 100%, whereas the specificity which, with the clinical criteria was 91.8%, rose to 99.6% with the presence of anti-RNP antibody. However, because testing of our criteria was made internally, they should be further tested, along with the other 2 sets of criteria by unrelated groups of clinical investigators, perhaps in a multicenter study.