Residual stress measurements on a single bead weld on a steel plate had been performed at the High Flux Reactor of the Joint Research Centre in 2003. For these measurements a relatively old diffractometer had been used. The results obtained were characterized by significant scatter of the data, and the measurement quality suffered from the short movement ranges of the specimen positioning table as well. In 2008, a second, nominally identical, specimen from the same activity was investigated on a second diffractometer, which allowed the repetition of the residual stress measurements using different measurement settings. The present paper compares the old and the new measurement results, with a view to assessing the impact of the instrumental settings on their quality. It has been found that the overall stress distributions from the 2003 and the 2008 measurements were in fact very similar. Nevertheless, the new settings used, such as in-situ specimen rocking, extended measurement duration, increased density of measurement positions and spatial resolution among others, have been found to have considerable impact on the real and on the apparent scatter of the experimental results showing that the presented methodologies can be used for improvement of neutron diffraction measurements.