There has been debate regarding the factor structure of the Children's Coping Strategies Scale (CCSS); in most previous studies there were different subscales, representing coping strategies. However, according to the theoretically multidimensional construct of coping, coping consists of an adaptive process and uses the lowest possible level of coping strategy. Accordingly, this study applies a new method of multidimensional item response theory (MIRT) to testing the latent structure of teenagers’ coping strategies and to improve the construct validity of the CCSS. The participants were 1,138 16-year-old high school students (48.2% female, 51.8% male) studying arts and science. This study compares three different model types: a unidimensional model, a simple multidimensional construct model, and a bifactor model. The results indicate that the latent structure of CCSS is not only described by specific coping strategies (e.g., rumination) but also by a general adaptation process, which is consistent with theoretical understandings of coping. Furthermore, the five-factor model, which contains a “reflection” dimension, is more suitable for Chinese teenagers.