Seeking light and transparent bridge designs, engineers and architects have found an efficient and artistic way to fulfill their requirements: steel tubular bridges. In these modern tubular truss bridges, welded K-joints have been shown to be critically susceptible to fatigue failure induced by repeated traffic loads. Despite the research devoted to the study of fatigue during the last 35 years, especially in the offshore oil industry, estimating their fatigue resistance of tubular truss bridges is difficult and solutions are approximate. This is due to the complexity of the parameters influencing bridge fatigue strength, including: 1) complex loadings, 2) applied stress concentrations in joints, 3) welding imperfections, 4) welding residual stresses, and 5) size effects. Some design rules and studies are available to estimate these effects; however, residual stresses in tubular K-joints remain unknown as well as their influence on fatigue crack propagation. Residual stresses are well known to have detrimental influence on fatigue crack propagation in other metallic structures, and further study is a necessary for improving the understanding of K-joint fatigue behaviour. To address this problem, the main objective of this research is to assess, experimentally and numerically, welding residual stresses and their influence on the fatigue response of K-joints. In this thesis, experimental measurements of the residual stress fields were investigated through neutron-diffraction, hole-drilling and X-ray methods. Results have shown that, principal residual stresses are oriented transversely, i.e. perpendicularly to the weld direction, both at and near the surface. Residual stresses can reach the S355 steel yield strength at the weld toe. This orientation is particularly detrimental for fatigue because it is also the orientation of the principal applied stresses, and as a consequence high residual and applied stresses superimpose. Moreover, it is proved that a restraining effect occurs in the gap region in-between the braces remaining the critical residual stresses high in the gap region. Two large scale tubular truss beams were subjected to high-cycle fatigue in order to assess the effect of residual stresses on fatigue response of welded tubular K-joints. These tests revealed that tensile residual stresses do not affect the fatigue crack growth in details loaded in tension (hot-spot 1), whereas they play a significant role in details loaded in compression (hot-spot 1c). Tensile residual stresses enable crack opening, making the fatigue cycle partly or entirely effective to crack propagation. Based on these tests and previous tests from the ICOM test database, it is recommended to design all (tension or compression) K-joints using a strength curve Srhs Srhs – Nf category 100 (for T = 16 mm). A thermo-mechanical finite element model was also developed to numerically reproduce the residual stress creation (using ABAQUS and MORFEO codes). Once the numerical results were validated by comparison with experimental results, the finite element model was used to model other K-joint geometries. A geometrical parametric study was performed to identify the most influential parameters affecting the residual stress distribution (technological size effect). It is found that a raise of the wall chord thickness T (proportional scaling) or of the wall thickness τ = t/T (non-proportional scaling) strongly increases the magnitude of tensile transverse residual stresses. Based on this parametric study, residual stress distributions with depth for transverse, longitudinal and radial directions are proposed. Finally, an extended finite element method (X-FEM) model was used to propagate a fatigue crack under residual and applied stresses. An analytical model based on the effective stress intensity factor range ΔKeff (fracture mechanics) and combined with the proposed distribution for residual stresses was also established to predict fatigue crack propagation in joints loaded in tension and in compression.