In this paper we tried to demonstrate the validity of C-Test using construct identification approach. In this approach to construct validation the factors which contribute to item difficulty are identified. The assumption is that the factors which make items difficult are actually the construct underlying the test. For the purposes of this study, 11 item-level and sentence-level factors, deemed to affect item difficulty, were entered into a regression analysis to predict classical item p-values. The 11 factors explained only 8% of the variance in item difficulties. This finding shows that lexical and sentential factors explain only a very small portion of the variance in p-values. It seems that a great amount of variation in item difficulties should be attributed to above-sentence and text level factors. The implications of the study for C-Test construct validity are discussed.