With the widely application of viscoelastic material in engineering structures, fractional calculus that characterizes its feature of hereditary and long-run memory associating with stochastic engineering systems becomes the hot issue. Specially, fractional PID controller is more attractive in theory and application. In this paper, fractional order PID (FOPID) control on reliability of stochastic dynamical systems subjected to Gaussian white-noise excitation is investigated. FOPID controller is mathematically approximated based on the characteristics of slow-varying process and the properties of generalized integral. After that, stochastic averaging method of energy envelope is applied to determine Backward Kolmogorov (BK) equation associated with conditional reliability function and Generalized Pontryagin (GP) equation associated with statistical moments of first-passage time. The numerical results illustrate that both the smaller fractional order α and greater fractional valueβ in PID controller can obtain better reliable results, besides, these results are in very agreement with the simulations from MonteCarlo method, which verifies the correction and efficiency of our proposed methods.