In order to study the structure reliability more exactly, the probability analysis method, instead of the traditional constant value analysis method, can be applied to overcome the uncertainty problem of the strength model. The reliability of railway vehicle bogie frame is influenced by many factors. In order to ensure the high reliability of bogie frame design, the precondition is found to be the dominating factor. In this paper, the parametric finite element model of bogie frame was established using the probability analysis method. The plate thicknesses, the service loads and material constants of bogie frame were set as the random variables and the distributed parameter values were obtained by statistics. The functions of failure state were established by allowed fatigue strength value and maximum principle stress of parent material or weld location of welded bogie frame. The reliability and parametric sensitivities of the random variables were simulated by Monte Carlo Method. Based on the analysis results, the random variables which are sensitive to the reliability of bogie frame can be controlled and optimized in practice so as to design high reliable bogie frame. The method applied in this paper provided an efficient way to evaluate the reliability and parametric sensitivity of the load bearing structure of high speed trains.