In the Eurocode system, for fatigue design of bridges, the recommended partial factor for fatigue traffic loads is set to 1. In this paper, the adequacy of this approach is investigated by performing a reliability analysis on two types of welded joint in a main girder of a steel motorway bridge. For this purpose, a weigh in motion measurement dataset belonging to a main Dutch motorway has been compared with the fatigue load model 4 of Eurocode EN 1991-2 with respect to the stress spectrum and the fatigue damage of two structural steel details. Several structural schemes have been considered to study the effect of the shape and length of the influence line. The distributions of the stochastic variables such as dynamic amplification, accuracy of the structural model, and future traffic trends have been estimated or taken from literature. Partial factors for fatigue loads have then been calibrated in such a way that the target reliability is obtained. The influence of each stochastic variable on partial factors has been studied by derivation of the sensitivity factors. The results show that a considerably higher fatigue partial factor is required for fatigue loads on road bridges than the value of 1 currently recommended in EN 1991-2.