The Remote Associates Test, generally used in creativity research, has Chinese versions for the three levels of "radical-word-vocabulary." However, research has not been conducted on the influence of the item components on the difficulties among these Chinese Remote Associates Tests (CRATs). The present study selected six item components from the construction rules for three versions of CRATs, collected 555 participants' responses to them, and analyzed the differences and similarities in the predictions of item difficulty among the six item components. The results show that item difficulty decreased if the targets are high-frequency or the stimuli are low-frequency words, but item difficulty increases if the stimulus can connect to more words, while the concreteness of the target does not significantly predict item difficulty. Moreover, the concreteness of the stimulus negatively predicts the difficulty of the CRRAT but positively predicts the difficulty of the CCRAT High association transfer increased the difficulty of the CWRAT but had no influence on the other two versions of CRAT. Therefore, these findings indicate the different influences of item components on item difficulty, strengthen the structural validity of the test, and also provide a reference for test development of the CRAT.