Aiming at the problem that high false positives in program static defect detection requires to be eliminated by lots of manpower, a correlation recognition method of program semantic defect warnings is proposed. By mining the deep correlation information between warnings, the warnings correlation is established, which helps to improve the manual judgment warning’s efficiency. In this paper, the symbolic expression and domain are used to represent the value of a variable, and the correlation derivation rule between warnings is established based on the logical relationship of the symbol expression. Then, in the defect detection phase, the warning is identified according to the defect trigger condition, and the correlation between different warnings is derived. Finally, the warning is automatically determined based on the correlation between the warnings. The test results of five actual C projects show that the proposed method can effectively identify the relationship between warnings and effectively reduce the work of manually determining warnings.