Steel tubular truss are lightweight, aesthetical and used in many different applications. These structures are sensitive to fatigue and specific verification methods have been developed for them, in particular by CIDECT [2001]. Due to assembly and welding, these structures contain large internal self-balanced stresses. Since tensile residual stresses tend to open cracks, they have a detrimental influence on the fatigue crack propagating under cyclic traffic loadings. Different studies on the behaviour of tubular structures have been focussing on the behaviour of tubular K-joints, one very much use type of joint, with tubes having low diameter to thickness ratios, for use in particular in bridge applications. In these cases, the CIDECT rule for reduction of fatigue strength with thickness becomes very penalizing and one has to see if alternatives are possible (changing the stress definition and/or the thickness effect formula). Also, many joints have members loaded under compression, such as upper chord joints in a truss. In these cases, tensile residual stress effect on fatigue life can be set apart from external loads effect. With welding imperfections acting like microcracks, residual stress can make the difference between a fatigue microcrack beginning to grow or not. Indeed, residual stress can raise the crack driving force above the threshold value. Moreover, it also influences the rate of crack propagation and therefore the residual stress distribution is essential to estimate accurately the crack development under fatigue loads. In this paper, work on the effects of size and residual stresses on fatigue life of tubular joints carried out at EPFL is summarised.