The equilibrium isotherms of the two enantiomers of phenylalanine anilide (PA) were measured by conventional frontal analysis at three different pH on a thermally-treated imprinted stationary phase selective for the L enantiomer. The first of these pH (buffer pH=3.0, pH(app)=4.0) is well below the apparent pKa (6.4) of the two solutes, the second (buffer pH=5.8, pH(app)=7.0) slightly below this pKa, and the third (buffer pH=7.0, pH(app)=8.3) well above it. The experimental data were fitted to several isotherm models. The best estimates of the parameters of these models are reported and discussed. The corresponding isotherms are compared with the experimental ones. The contributions of the enantioselective and nonselective interactions could be separated. The results obtained show that the saturation capacity is always smaller for D-PA than for L-PA, the template. The analytical separation is best at pH=3.0 because of a good separation factor (2.82) and short retention times. A good compromise between the resolution and the saturation capacity is obtained at pH=5.8, for which the best preparative separation is found. Both analytical and preparative results are poor at pH=7.0 because the separation factor is low (1.32). At this pH, the isotherm remains nearly linear in the whole concentration range accessible to measurements. The number of nonselective sites increases with increasing mobile phase pH slightly faster than the number of selective sites, suggesting different pKa ranges for the two type of sites. Moreover, the binding energy and the homogeneity of the selective sites decreases with increasing pH. These results agree with a binding site model involving more than one carboxylic acid group, providing charge complementarity and hydrogen bond donors for binding of L-PA.