The weldability of stainless steels is largely controlled by the chemical composition, and alloys with ferritic or ferritic-austenitic solidification show the highest resistance to hot cracking. As the resulting phase balance also affects the final properties, it may be beneficial to both foresee and measure the weld metal ferrite content. The WRC ‘92 constitution diagram is currently the most accurate prediction tool available, but it does not take the cooling rate into consideration and the precision may be less accurate for stainless steels with high ferrite numbers (FNs). This study aims to assess the reliability of the WRC ‘92 diagram for weld metals with FN > 50. The chemical composition was altered through gas tungsten arc welding (GTAW) of UNS S32205 with ER347 filler wire that had been coated using physical vapor deposition (PVD) with either niobium (Nb), copper (Cu), nickel (Ni), manganese (Mn), carbon (C), or silicon (Si). The actual ferrite content was evaluated using image analysis, FeriteScope and X-ray diffraction (XRD). While predictions from the WRC ‘92 diagram were deemed acceptable for Ni, Si, and Mn, notable deviations were observed for Nb, Cu, and C. The FeriteScope exhibited a consistent trend with image analysis, albeit with slightly higher FN values, wider scatter, and the conversion factor from FN to vol% is open for discussion. The lowest accuracy and largest spread were obtained using non-contact XRD, rendering it unsuitable for ferrite measurements of welds. These findings underscore the need for improved prediction tools and appropriate measurement methods for assessing ferrite content in duplex weld metals.