<p>Water cut is a critical parameter in reservoir evaluation because it strongly influences residual oil estimates. Conventional water cut calculation methods often lack effectiveness in complex reservoirs, leading to low prediction accuracy. This study addresses the Triassic Chang 8 Formation in the Hongde region by proposing an empirical water cut prediction method that uses nuclear magnetic resonance (NMR) logging to improve precision. The methodology comprises several steps: first, combining NMR measurements and relative permeability experiments on 20 core samples clarified the impact of pore structure on relative permeability, and an NMR <i>T</i>₂-derived parameter was defined to quantify pore structure. Subsequently, the target formation was classified into three formation types, and a relative permeability prediction model was developed for each type using NMR logging. Following this, models were established to estimate water saturation, irreducible water saturation, and irreducible oil saturation so these inputs could be derived from conventional well logging data. Finally, a formula for calculating reservoir water cut based on Darcy’s law was derived and applied to field data in the target reservoirs to predict water cut continuously. Comparison of predicted water cut values with drill stem test data demonstrated the efficacy of the proposed model, achieving an accuracy of 83.3%. This methodology holds promise for optimizing oilfield production development with reliable outcomes.</p>