Rock surface luminescence dating has developed into a promising approach to constrain depositional ages of clasts. The prerequisites for its successful application are the precise determination of the clast-specific dosimetry and the sufficient pre-burial resetting of the luminescence signal into a certain depth beneath the clast-surface. Dosimetry of clasts is significantly influenced by the internal beta dose rate, which largely depends on grain-size range and internal potassium content of feldspar within rock slices. In the current study, six clasts are investigated from three distinct fluvial terraces of the Bruche River in north-eastern France. Previously published independent age control from sandy deposits of the terraces indicate depositional ages of ~12–14 ka for the youngest, ~27–35 ka for the middle, and a minimum age of ~200 ka for the oldest terrace. Clast-based multi-elevated-temperature post-IR infrared stimulated luminescence (pIR) ages show systematic deviations from age control, overestimating by ~50-150% for the youngest terrace, ~10-30% for the middle terrace, but confirming the minimum age for the oldest terrace. The present study provides methodological advancements by estimating grain-size range and K-content in feldspar grains within rock slices using µ-X-ray fluorescence and microprobe analysis, respectively. Modern clasts from the present riverbed indicate that high-temperature pIR signals retain significant residual ages in rock slices, ranging from ~2-3 ka for pIR110 over ~5-6 ka for pIR170 to ~8 ka for pIR225 signals, while IR50 residual ages remain negligible (<1 ka). This study also demonstrates the potential for using fully saturated rock slices from clast interiors to perform reliable fading correction. As the clast-based luminescence ages overestimate the corresponding ages derived from sandy deposits, this highlights the remaining limitations that must be considered for this approach.