In theory, topical delivery has substantial potential to treat local and some systemic disease states more effectively than systemic delivery. Unfortunately many, if not most, drug candidates for topical delivery lack the requisite physicochemical properties that would allow them to permeate the skin to a clinically useful extent. One way to overcome this obstacle to effective topical delivery is to make a transient derivative of the drug, a prodrug, with the correct physicochemical properties. But what are those correct properties and can the directives for the design of prodrugs be applied to the design of new drugs, their analogs or homologs? For some time increasing the lipid solubility (S (LIPID)) or its surrogate, the partition coefficient between a lipid (LIPID) and water (AQ) (K (LIPID:AQ)), has been the standard working paradigm for increasing permeation of the skin, and the permeability coefficient (P = distance/time) has been the quantitative measure of the result. However, even the earliest reports on non-prodrugs such as alcohols showed that working paradigm was incorrect and that P should not be the relevant measure of permeation. The shorter chain and more water soluble alcohols exhibiting lower K (LIPID:AQ) values gave the greater flux values (J = amount/area x time; the more clinically relevant measure of permeation), regardless of whether they were applied neat or in an aqueous vehicle, while P showed opposite trends for the two applications. Subsequently a large volume of work has shown that, for prodrugs and non-prodrug homologs or analogs alike, S (AQ) (not solubility in the vehicle, S (VEH)) as well as S (LIPID) should be optimized to give maximum flux from any vehicle, J (MVEH): a new working paradigm. The dependence of J (MVEH) on S (AQ) is independent of the vehicle so that S (AQ) as well as S (LIPID) are descriptors of the solubilizing capacity of the skin or S (M1) in Fick's law. The inverse dependence of J (or P) on molecular weight (MW) or volume (MV) remains. Here we review the literature that leads to the conclusion that a new working paradigm is necessary to explain the experimental data, and argue for its use in the design of new prodrugs or in the selection of candidate analogs or homologs for commercialization.