Abstract: Discusses the importance of correlation coefficients between random space variables in the reliability analysis and the reliability-based sensitivity of those correlation coefficients. Based on the MPPPM for the reliability analysis where the basic random variables are non-independent with each other and can be processed directly, the analysis of the importance and sensitivity mentioned above are carried out. Then, the effect of changing correlation coefficient on the structural reliability is obtained by the reliability-based sensitivity analysis of correlation coefficients. A numerical example is given to show how the method proposed is applied to practical project. The results illustrate the importance of correlation coefficients in reliability analysis and how it affects the structural reliability, thus providing a reference beneficial to statistics especially the reliability design.