Abstract: This paper analyzed the characteristics and applicable scenarios of four software reliability growth models (SRGM), such as G-O, Delayed S-shaped, Ohba- Chou and P-N-Z, studied the reliability growth trend analysis method and evaluation index of software failure data set, put forward an optimization algorithm of SRGM parameter calculation, designed and implemented a SRGM analysis tool (SRGM Tool) on the basis of theoretical research. Taking the failure data set recorded by railway interlocking software as an example, the paper studied the application method of railway SRGM, with the help of SRGM Tool, used four models to evaluate the reliability of railway interlocking software, and according to the experimental results, analyzed and determined a reliability growth model suitable for the software, which verified the effectiveness of the method, and could provide reference for reliability analysis and fault prediction of railway software.