Objective To explicate whether mixed connective tissue disease (MCTD) is a distinct disease and evaluate the reliability of three different diagnostic criteria proposed by Sharp, Alarcon-Segovia and Kasukawa respectively.Methods Clinical follows-up of 50 MCTD patients lasted 2-8 years (80% >5 years). HLA-A, -B as well as -DR typing was performed by complemently dependent cytotocity assay. Autoantibody profile was detected by counterimmune electrophoresis (CIE).Results Thirteen (26.0%) of the 50 MCTD patients subsequently developed other connective tissue disease (OCTD), including 7 systemic lupus erythematosis (SLE), and 6 progressive systemic scleroderma (PSS). Among 23 of the MCTD patients fulfilling Sharp's criteria, 1(4.3%) developed PSS, but among 23 of the patients fulfilling Kasukawa's, not Sharp's, 7(30.4%) developed OCTD and among 27 of the patients fulfilling Alarcon-Segovia's, not Sharp's, 12(44.4%) developed OCTD. In the frequencies of DR4 and DR5, there were significant differences between patients fulfilling Sharp's (60.9%, 56.5%) and controls (24.3%, P<0.005, RR=4.7 and 21.4%, P<0.005, RR=4.6%), but there were no significant differences between the patients not fulfilling Sharp's and normal control (P>0.05).Conclusions MCTD is a distict rheumatic disease. Sharp's criteria is the most reliable for diagnosis of MCTD.