Purpose <p>The purpose of this study was to investigate the theoretical and <i>in vitro</i> experimental prediction of food effects on oral drug absorption, focusing on solubility-epithelial membrane permeation-limited cases (SL-E).</p> Method <p>Bosentan, fidaxomicin, pranlukast, and rifaximin were employed as model SL-E drugs. Celecoxib and danazol were employed as solubility-unstirred water layer permeation-limited cases (SL-U) for comparison. Theoretical predictions of food effects were based on the rate-limiting steps of the fraction of a dose absorbed (<i>Fa</i>) (FaRLS) (Fa rate-limiting step). μFLUX was used as a dissolution-permeation flux (<i>J</i>_{<i>μFLUX</i>}) experiment. Fasted and fed state simulated intestinal fluids (FaSSIF and FeSSIF, respectively) were employed as the donor solution.</p> Results <p>For all SL-E drugs, the food effect on <i>Fa</i> was theoretically predicted to be 1.2, irrespective of bile micelle solubilization (FaSSIF/FeSSIF: bosentan (2.1), fidaxomicin (2.3), pranlukast (9.1), and rifaximin (3.5)). Theoretically, an increase in solubility by bile micelles is counterbalanced by a decrease in effective permeability (<i>P</i>_<i>eff</i>) due to a decrease in the free fraction (<i>P</i>_<i>eff</i> is defined based on unbound + bound drug concentration (<i>C</i>_<i>D</i>)). This prediction was consistent with the clinical data (fed/fasted AUC ratio: 1.1, 1.0, 1.3, and 1.6, respectively). In μFlux, even though <i>C</i>_<i>D</i> was markedly higher in FeSSIF than in FaSSIF (1.9, 3.1, 20, and 3.3-fold, respectively), <i>J</i>_{<i>μFLUX</i>} was less enhanced (0.91, 0.81, 2.4, and 0.81-fold, respectively). For the SL-U drugs, as theoretically expected, <i>J</i>_{<i>μFLUX</i>} was increased as <i>C</i>_<i>D</i> was increased, which was consistent with the clinical data.</p> Conclusion <p>FaRLS appropriately predicted the food effect for the SL-E drugs. The mechanism was experimentally confirmed by μFlux.</p>