Induced seismicity in Groningen, Netherlands, has necessitated the assessment of a building stock not originally designed for seismic hazards. With no prior national standards for seismic assessment, early evaluations relied on Eurocode EN 1998–3, which was not specifically suited for the region’s shallow earthquakes and its unique building stock, mostly composed of low-rise unreinforced masonry (URM) buildings with cavity walls, slender piers, and large openings. Since 2015, the Dutch NPR 9998 guidelines have introduced various assessment methods, but inconsistencies in assessment outcomes remain a challenge. A full-scale URM house test at TU Delft, linked to a blind prediction contest, highlighted significant variability in key response parameters, such as base shear and displacement capacity. This paper discusses three initiatives aimed at improving consistency in seismic assessment. First, cross-comparison exercises evaluated different nonlinear dynamic and static procedures on reference buildings. Second, a semi-probabilistic typology-based approach was developed for global seismic assessment, with a similar method under development for local out-of-plane URM wall assessments. Third, an experimental campaign on URM gable wall failures, combined with a blind prediction contest, seeks to enhance consistency in computational approaches for this under-researched typology. While progress has been made in improving the consistency of assessments, challenges remain, particularly in accounting for local failure mechanisms. The semi-probabilistic approach offers a promising path forward, but further refinements are needed for broader application beyond the Dutch context.