The aim of this study was to determine the skin penetration of benzophenone-3 in vitro and in vivo in order to investigate a possible influence of formulation. Six different vehicles, three solvents and three different emulsion types were evaluated in vitro and in vivo. Each vehicle was applied to the skin model at 2 mg cm(-2). First, histological studies on ear pigskin and human skin were evaluated. In vitro measurements were performed with static diffusion cells using pigskin at 1, 2, 4, and 8-h. In vivo, benzophenone-3 concentration in stratum corneum was evaluated by the stripping method after 30-min application on forearm of volunteers. It was shown that ear pigskin and human skin appear similar and in both experiments significant differences between vehicles were noticed. The six vehicles could be ranked in the same order of benzophenone-3 skin concentration. At 8-h, the highest concentration of benzophenone-3 in skin was obtained with propylene glycol, and O/W submicron emulsion. On the contrary. the two oily solvents. W/O emulsion and O/W coarse emulsion restrain the concentration of this UV-filter in the skin. At each time, permeability in vitro and in vivo were well correlated. Low concentrations were measured in the receptor fluid suggesting that percutaneous absorption of this UV-filter across the skin would be minimal. The in vitro and in vivo skin penetration capacity of benzophenone-3 from six vehicles was confirmed and quantified. A satisfactory relationship between binary in vitro and in vivo was established.