Iontophoresis of the dipeptide tyrosine-phenylalanine (TyrPhe), the protected amino acid tyrosine- β -naphthylamide (Tyr- β -NA) and the glucose derivative benzyl-2-acetamido-2-deoxy-α-D-glucopyranoside (BAd-α-Glc) used as model compounds was investigated in human epidermis in vitro at pH 3 and pH 4.5 under constant voltage application, the purpose being to delineate the contribution of tissue retention, lipophilicity, ionic valence and convective transport of these compounds to iontophoretic enhancement in a quantitative fashion and ultimately gain an improved mechanistic understanding of iontophoresis. Retention of BAd-α-Glc in epidermal tissue during permeation was considerable and was reduced upon iontophoresis in the presence of extraneous tyrosine (Tyr) and phenylalanine (Phe) added to the donor solution, this evoking an increase of the apparent iontophoretic permeation. This suggests firstly, the occurrence of an interaction between BAd-α-Glc, Tyr and Phe impacting iontophoretic enhancement and secondly, the potential of using amino acids as adjuvants to modulate iontophoresis of selected compounds. This influence of tissue retention on iontophoretic permeation in the presence of Tyr and Phe was not observed for TyrPhe or Tyr-β-NA or when BAd-α-Glc, TyrPhe and Tyr-β-NA were used concomitantly rather than individually in the absence of Tyr and Phe. The effect of lipophilicity, ionic valence in the aqueous permeation pathway and convective transport by electroosmosis of the three permeants on iontophoresis was assessed simultaneously for all permeants by analyzing the experimental data using a theoretically derived model for iontophoretic enhancement that encompassed these factors. This model was based on an extension of the modified Nernst-Planck equation, whereas the difference of ionic valence between the aqueous domain of tissue and the bulk solution was evaluated on the basis of a pH shift due to the electrical double layer at the lipid/aqueous interface in the epidermis using the Poisson-Boltzmann equation. Using deduced parameter values characterizing the involved factors, a very good agreement between model calculated and experimental enhancement data was obtained. At pH 4.5, a very weak convective transport due to electroosmosis from cathode to anode was evident, indicating an isoelectric point of the epidermis slightly above 4.5. At pH 3, electroosmotic transport was approximately 10-fold stronger than at pH 4.5 which reflected a high positive surface charge density of the epidermis at pH 3 and a low one at pH 4.5. The pH in the aqueous permeation pathway, estimated to differ from that of the bulk in accordance with these charge densities, produced an ionic valence in this pathway dependent on pK a that accounted consistently for the flux due to the direct effect of the electric field on the ionic permeants. The ratio of lipid to aqueous pathway passive permeability coefficient was