The purposes of this paper were to investigate the construct validity of Test of Proficiency in Korean(TOPIK) and to identify the adequacy of item types (subjective / objective) in Writing Section. In order to achieve the goals, the test data from three administrations of TOPIK(22th, 23th, and 24th test) were analyzed. Exploratory factor analyses and correlation analyses were conducted to obtain empirical evidences. The major findings of this study were two-folded: First, each of 4 sections in TOPIK did not represent unique domain as designed by test specification. The correlations among scores of 4 sections were relatively high compared with the results of TOEIC which shared similar test purpose with TOPIK. Also it was found that basic level test form tended to have unidimensional characteristic rather than advanced level test form. Second, there was not much unique effect of subjective items in Writing section compared with objective items. Correlations between subscores of subjective items and scores of Writing section became lower in the advanced form. The limitations and practical implications of the study to improve TOPIK were also discussed.