The presence of residual stresses is a determining factor in the buckling behaviour of steel beams. It is therefore necessary to know the magnitude and the distribution of residual stresses in welded beams. Experimental measurements of these residual stresses are expensive, timeconsuming and they give no insight in the evolution of the stresses during the welding process. For these reasons it is interesting to have a numerical model which allows a prediction of the residual stresses due to welding and which allows a study of the parameters that have an influence on the residual stresses (e.g. yield stress, heat-input, thermal expansion coefficient, …). Such a numerical model can result in a proposal for a welding process that leads to less detrimental residual stress distributions in welded I-beams. However, the results of the numerical model must always be checked with reality. This contribution contains the experimental verification of the finite element model for both single plates and I-beams with rather small dimensions. First, an overview of generally accepted influencing factors on the residual stresses is given. After this introduction, a setup which allows control of the parameters for making reproducible weld seams is presented. This setup is usable for both plates and I-beams. Next it is described how the residual stresses in a plate and a beam were measured. Finally the results obtained from different configurations in the experiments are compared with the results from the FE-model. It is shown that the FE-model gives reliable predictions for the residual stresses.