Plasma concentration profiles of arundic acid ((R)-(−)-2-propyloctanoic acid), an oil-like medicine, administered as soft-gel capsules in human clinical tests were predicted from the dissolution test data of the soft-gel capsules with different storage terms (short- and long-term stored drugs) by applying the in vitro–in vivo correlation (IVIVC). We established two linear-regression IVIVCs, which were characterized by either the in vitro dissolution behaviors against the pH 8.0 dissolution medium or the pH 6.8 dissolution medium containing 2% sodium dodecyl sulfate (SDS), in this study. Also, the prediction accuracies for the in vivo plasma profiles in humans for these two IVIVCs were compared. Regarding dissolution from the long-term stored capsule in pH 8.0 dissolution medium without surfactant, the prediction accuracies of the in vivo plasma profiles in humans were not satisfactory for the obtained IVIVC. The use of pH 6.8 dissolution medium containing 2% SDS, according to the Japanese guideline, improved the dissolution of the long-term stored capsule. Furthermore, the prediction accuracies for the in vivo plasma profiles in humans for these two IVIVCs were compared. The IVIVC established by the in vitro dissolution data obtained with the dissolution medium containing surfactant more effectively predicted the plasma drug concentration profiles following oral administrations of the soft-gel capsules under both storage conditions.