Our group has been focused on crack propagation in the high-cycle fatigue regime in order to develop reliable numerical predictions of residual fatigue life (RFL) of engineering components. In this contribution, parameters that influence RFL of railway axles in the most significant way were identified. In particular, it was the residual stress field induced by surface heat treatment and the long crack threshold. These parameters were found to be exceptionally important for investigation and for application of a probabilistic approach. There are two reasons why these parameters were selected and identified as more relevant than the others. First, a precise determination of these parameters is one of the most difficult tasks due experimental inaccessibility to direct measurement. The use of indirect methods or methods influenced by other factors is often the only option. This results in large measurement uncertainties, which may in some cases lead to non-conservative predictions. At the same time, the actual values of threshold and residual stress are naturally scattered due to stochastic phenomena. Second, the analysis of fatigue crack growth rates and thresholds revealed that the effect of these factors on RFL is much more significant than the traditionally studied material parameters, such as yield stress or tensile stress. Our methodology is demonstrated on the example of a probabilistic distribution of the results obtained by the Monte Carlo method with the consideration of stochastic variation of the inputs, namely the threshold, the Paris law constant (NASGRO) and scale parameter (Castillo et al. model), the magnitude of the residual stress field and the stress field induced by press fitting. The axle was loaded by a characteristic operational spectrum. The results were compared to the 1:1 pieces of railway axles tested under the same loading. The experimental RFLs were close to the median of the statistical distribution of the numerically simulated RFLs.