The geometric parameters and materials of the building as well as the loading in reality are uncertain quantities. Therefore, for the design of the project to more accurately reflect the actual working characteristics of the structure, structural reliability has attracted researchers recently. In this article, the author presented the optimization problem of nonlinear steel trusses to consider the conditions of the structural failure probability. The objective function is the total weight of the truss, while the design variables are the continuous variables in given value ranges. Constraints corresponding to the strength and serviceability load combinations. The structural failure probability is determined using “Latinized” Partially Stratified Sampling (LPSS) and Effective Importance Sampling) (EIS) and is named LPSS-EIS. The optimal process is performed using an improved differential evolution (DE). A 72-bar steel truss is studied to illustrate the optimization problem developed.