This thesis work intended to verify an analysis method for welded, thin-walled geometries. Guidelines for stress evaluation in welded structures exist and are standardised, but they are often verified for structures with higher plate thicknesses, such as those found in the offshore industry. Thinner structures are commonly analysed using the hotspot method, but questions still exist wherever the method is valid and can provide conservative results in thin-walled geometries.  One goal of the thesis work was to create a test plan to experimentally verify the results given by FE models of welded structures, as well as to investigate the strain gradient close to the weld toe. The plan, as well as two different welded specimens were designed and manufactured on which future analysis can be performed.  The hotspot method was also evaluated using FE analysis on geometries where two tubes were welded together with a T-joint with varying diameter, thickness and applied loads. A total of 13 different models were created using solid elements where hotspot stress extrapolation was evaluated using different extrapolation points and evaluation paths. In conclusion, it was found that the method provides a correct extrapolation of the geometric stress when stress extrapolation points at a distance of 0,4t and 1,0t from the weld toe are used (t=plate thickness). It was also found through the analysis that the geometric stress can be harder to differentiate from the non-linear part of the stress gradient for some profiles with a thickness of 0,89 mm. In some cases, this resulted in a small part of the non-linear stress being included in the extrapolation which increased the extrapolated hotspot stress. Comparisons between the hotspot stress and geometric parameters showed that stress concentration factors can be created which reduce the need for time-consuming FE models.