The key aspects of the pharmacokinetics of transdermal delivery systems including time lag, steady-state plasma levels and decline phase are illustrated in this review. The 7 currently marketed transdermal systems [nitroglycerin (glyceryl trinitrate), estradiol, clonidine, fentanyl, nicotine, scopolamine (hyoscine) and estradiol/norethisterone acetate] are discussed, as are systems in development. Single-dose absolute bioavailability studies characterise the period of onset, the steady-state plateau and the declining phase, and typify transdermal delivery. More complex temporal profiles result from interactions with enhancers or removal of the system before steady-state conditions are achieved. Clinically these systems are used to achieve multiple peak serum estradiol concentrations after application of transdermal estradiol, and an initial peak systemic concentration of testosterone after application of transdermal testosterone. Multiple-dose, dose proportionality and skin site bioequivalence studies are needed for the full pharmacokinetic characterisation of a transdermal delivery system. The relationship of system design to variability is discussed. Although the data are limited, population factors, cutaneous metabolism and tolerance all appear to influence the disposition of drugs administered transdermally. For example, the route of delivery influences which nitroglycerin metabolite predominates. Furthermore, as a result of tolerance to nitrates, a transdermal delivery system must be removed for 8 to 12 hours for optimal effect. Therefore, transdermal delivery systems, designed on the basis of pharmacokinetic principles and concentration-effect relationships, have the potential to provide optimal therapy for the treatment of some conditions.