In a previous study on the human mass balance of DS-1971a, a selective NaV1.7 inhibitor, its CYP2C8-dependent metabolite M1 was identified as a human disproportionate metabolite. The present study assessed the usefulness of pharmacokinetic evaluation in chimeric mice grafted with human hepatocytes (PXB-mice) and physiologically based pharmacokinetic (PBPK) simulation of M1. After oral administration of radiolabeled DS-1971a, the most abundant metabolite in the plasma, urine, and feces of PXB-mice was M1, while those of control SCID mice were aldehyde oxidase-related metabolites including M4, suggesting a drastic difference in the metabolism between these mouse strains. From a qualitative perspective, the metabolite profile observed in PXB-mice was remarkably similar to that in humans, but the quantitative evaluation indicated that the area under the plasma concentration-time curve (AUC) ratio of M1 to DS-1971a (M1/P ratio) was approximately only half of that in humans. A PXB-mouse-derived PBPK model was then constructed to achieve a more accurate prediction, giving an M1/P ratio (1.3) closer to that in humans (1.6) than the observed value in PXB-mice (0.69). In addition, simulated maximum plasma concentration and AUC values of M1 (3429 ng/ml and 17,116 ng·h/ml, respectively) were similar to those in humans (3180 ng/ml and 18,400 ng·h/ml, respectively). These results suggest that PBPK modeling incorporating pharmacokinetic parameters obtained with PXB-mice is useful for quantitatively predicting exposure to human disproportionate metabolites. SIGNIFICANCE STATEMENT: The quantitative prediction of human disproportionate metabolites remains challenging. This paper reports on a successful case study on the practical estimation of exposure (C max and AUC) to DS-1971a and its CYP2C8-dependent, human disproportionate metabolite M1, by PBPK simulation utilizing pharmacokinetic parameters obtained from PXB-mice and in vitro kinetics in human liver fractions. This work adds to the growing knowledge regarding metabolite exposure estimation by static and dynamic models.