Bogie frames of electric cars play an important role as a structural member for the support of vehicle loading. To guarantee more than 25 years' durability, a lot of study has been carried out for the prediction of the fatigue life of the bogie frame in experimental and theoretical domains. One of the new methods being applied in recent years for rolling stock structures is a reliability-based approach. In this paper, using one of these methods, we estimate fatigue life of the bogie frame of an electric car, which was developed by the Korea Railroad Research Institute. First of all, we have carried out tensile fatigue tests with several types of specimens: as-welded, ground on weld toe, reinforcement-removed and post weld heat treated specimens for each type aforementioned. In addition, we perform the probabilistic distribution tests of the S-N curves for the ground specimens. Normal, lognormal and Weibull distributions are used to model the distributions. Along with the obtained S-N relations, we use two approaches to evaluate the fatigue life of the bogie frame. In the first approach probabilistic distribution of the S-N curve is used. The S-N curves are obtained in function of failure percent from the specimen tests. Then the fatigue life is estimated by use of Miner-Palmgren hypothesis. In the second approach, load spectra measured by strain gauges are approximated by a two variable Weibull distribution. And a limit state function for the criterion of failure is derived by use of Miner-Palmgren hypothesis. The fatigue life is evaluated by a reliability engineering method.