An EB welded P91 steel plate is selected to measure longitudinal residual stresses (acting in the welding direction) using the contour method. The EB weld causes short length scale residual stresses within the narrow width of the fusion zone with steep stress gradients lateral to the weld line. In order to capture the short length scale residual stresses and improve the spatial resolution of the contour method, the special measures were taken for all main four steps of the technique. For this improved approach, first the specimen cutting was performed with the special care. A thin cutting wire diameter (10 µm rather than standard 25 µm) was used to achieve a better surface roughness, the surface deformation was sampled at a higher frequency, and then, the data analysis steps were performed precisely to measure the short length scale residual stresses, the data smoothed over an optimum length-scale and an appropriate finite element mesh density was chosen. The success of the measurement is assessed by comparing the new contour method result to the published first conventional contour measurement result and the neutron diffraction result for the same component. Previously, the initial conventional contour method measurement was unable to resolve the tensile stress peaks in the HAZs adjacent to the weld fusion zone. However, the new contour measurement residual stresses are in good agreement to the neutron diffraction results. It is also shown that the new contour measurement results has been successfully captured the steep stress gradient on both sides of the weld centre-line (0 mm) with tensile stress peaks situated just 2.6 mm apart.