In this chapter, based on the steepest descent sensitivity vector, the formulation of first-order reliability method is disscused, the improved vesions of the FORM formulated by steepest descent iterative snesetivety vector are collected that these improvd version of steepest decent formulation are enhanced for robustanss using a step size strategy. Four improved HL-RF methods as well-known steepest descent method are developed including (i) improve HL-RF, (ii) chaos control method, (iii) relaxed HL-RF, (iv) Non-negative constraint method that they are formulated for improving gradient line search direction based on a step size. Also, several cast studies are provided for comparing the efficiency and robustanss of these methods.