Transdermal iontophoresis offers a non-invasive sampling method for therapeutic drug monitoring. This study examined whether iontophoretic extraction (a) is concentration dependent, (b) reflects the subdermal level of unbound drug, (c) follows protein binding changes, and (d) becomes truly non-invasive when a co-extracted compound is used as an internal standard for calibration. Iontophoresis was conducted in vitro using dermatomed pig-ear skin. The subdermal solution was a buffer containing phenytoin at therapeutic concentrations, an internal standard at fixed level, human albumin and/or valproic acid. The ionized form of phenytoin was recovered at the anode by electro-migration, while the neutral form was extracted to the cathode by electroosmosis. A satisfactory correlation between the reverse iontophoretic extracted amount of phenytoin and the subdermal concentration was observed. Iontophoresis extracted only the free fraction of phenytoin. At steady state, reverse iontophoresis monitored changes in free drug concentration provoked in the subdermal compartment. Acetate was introduced at a fixed concentration into the subdermal compartment to act as an 'internal standard'. Subsequently, acetate and the ionized form of phenytoin were co-extracted to the anode. The ratio of the extracted amounts was proportional to the subdermal concentration ratio demonstrating a means by which the method may become truly non-invasive.