Approximately 80% of chemical products used in daily life involve heterogeneous catalytic reactions. How to shorten the catalyst development cycle and efficiently screen catalysts with the superior performance has become a challenge. Among various types of catalysts, two-dimensional materials have attracted recent attention due to their unique structure and electronic properties. It is thus necessary for two-dimensional materials with some specific reaction requirements to clarify the relationship between catalyst structure and function. Descriptors can correlate the structural properties, electronic properties and performance of catalysts, and the structure-activity relationship of catalysts can be revealed through descriptors. The principle of each descriptor?s association with catalyst performance and its mechanism were introduced. Structural descriptors have a better performance in predicting the activity of two-dimensional carbon materials. Compared with the structural descriptors, electronic descriptors can match the parameters of catalytic performance. It is possible to develop the binary descriptors that combine the merits of both structural and electronic descriptors. In the era of big data, the development of descriptors is gradually integrated with emerging methods such as machine learning, and some performance prediction models for multi-dimensional parameters need to be developed. With the progress of artificial intelligence technology, big data-driven performance descriptors will facilitate the development of 2D material catalysts.