Abstract: Liquid core reduction (LCR) technology, originally developed for continuous thin-slab casting, allows space for a submerged entry nozzle in a mold while improving production efficiency. Recent experimental attempts explore the implementation of LCR in regular slab casting processes. However, regular slabs (2–3 times thicker than thin slabs) face critical challenges in terms of excessive deformation and stress concentration under external forces, which induce intermediate cracks and thus hinder successful LCR adoption in regular slab production. This study evaluates the feasibility of LCR for producing regular slabs and identifies optimal reduction parameters to prevent crack initiation. A three-dimensional thermal–mechanical coupled model is proposed using the finite element method (FEM), integrated with the equivalent replacement liquid steel (ERLS) method and the normalized Cockcroft–Latham damage model, to achieve quantitative prediction of intermediate crack risk during the LCR process. The ERLS model simulates the extrusion flow and expulsion behavior of the liquid core, and its accuracy is validated against actual production measurements. To identify the critical damage value leading to intermediate crack initiation, this study conducts a consistency analysis between high-temperature tensile tests and FEM-based simulations using damage models. Based on this value, crack prediction is performed for Q355 slabs with cross-sectional dimensions of 170 mm × 1450 mm. Using the prediction results, an optimal reduction scheme is determined, wherein the second segment accounts for 50% of the total reduction, the third segment for 32.5%, and the fourth segment for 17.5%, with the theoretical value of maximum reduction being 34 mm. These results provide actionable guidelines for the potential implementation of LCR in regular slab-casting systems.