The focus of this paper is on the use of first order reliability method (FORM) for reliability analysis of soil liquefaction potential. First, an empirical equation for determining liquefaction resistance, in terms of cycle resistance ratio (CRR), based on cone penetration test, is established through "neural network learning" of case histories. This CRR model is established based on cyclic stress ratio (CSR) model proposed by Idriss and Boulanger, and together, they form a liquefaction limit state. Within the framework of the FORM, the uncertainty of this limit state model is estimated by means of Bayesian mapping functions. Accurate reliability index and thus the probability of liquefaction can be calculated using FORM by considering both model and parameter uncertainties.