Welded circular hollow section (CHS) joints typical to bridge construction have been examined in an experimental, numerical, and analytical study with respect to their fatigue strength. Evidence from this study has indicated lower fatigue strength for joints with thicker failed members. A re-examination of the data on which the size effect correction for welded tubular joints in current specifications is based has raised questions about its validity, especially for joints with greater wall thicknesses. Subsequently, fatigue lives of a range of welded CHS K-joints have been calculated based on linear elastic fracture mechanics (LEFM) and the finite element method. Results show that the relative fatigue strength of these joints as affected by changes in wall thickness, joint scale and loading produce a so-called size correction that is substantially different than that in the current specifications. A more coherent definition of the size effect, whereby the influence of different joint parameters and dimensions is accounted for individually, is suggested.