This study intended to develop the ”Modeling Ability Analytic Index (MAAI)” for assessing students' modeling ability at different stages, and planed to make teachers understand the meaning of modeling well. In the application of index, not only can MAAI be used to design modeling assessment tools, but also it can be used to design corresponding coding scheme to investigate students' modeling abilities. In our empirical study, we used the topic of electrochemistry to analyze 79 senior high school students' modeling abilities after they took the modeling test. The research findings suggest that: (1) Students' modeling ability was categorized low. The result was almost the same as Hestenes' research (1992). (2) The students’ modeling ability had significant correlation with their achievement in science subjects and academic achievement. (3) For low-achieving students, their modeling ability had no significant correlation with their academic achievement, but had significant correlation with their achievement in science subjects. Therefore, we considered that modeling ability is an important foundation of learning science. If teachers use MAAI more often, than it will help students learn science better.