<p>The fatigue strength of a welded joint was investigated using full thickness (50 mm) and reduced thickness (5 mm) specimens, keeping the surface of the thicker specimens, including the weld cap, with the objective to determine the fatigue strength of thick members using smaller specimens. Initially, the fatigue test results of the reduced thickness specimens were significantly scattered. The experimental scatter could not be explained by the geometry of the weld toes: all the specimens were laser scanned prior to testing but no significant geometrical differences were found at the weld toe. However, the misalignment had a significant impact on the test results for the thin samples. The results show that, using a correction for the misalignment of reduced thickness specimens, it is possible to reduce significantly the scatter of the experimental results. Furthermore, the results indicate that the fatigue strength of the reduced thickness specimens is very close to the fatigue strength of the full thickness specimens. In conclusion, the results question the thickness correction for the fatigue strength in high quality butt welds produced by the submerged arc welding process.</p>