In vitro and in vivo skin absorption of the pesticide propoxur (2-isopropoxyphenyl N-methyl carbamate, commercially Baygon(TM) and Unden (TM); log Po/w 1.56, MW 209.2) was investigated. In vivo studies were performed in rats and human volunteers, applying the test compound to the dorsal skin and the volar aspect of the forearm, respectively. In vitro experiments were carried out in static diffusion cells using viable full-thickness skin membranes (rat and human), non-viable epidermal membranes (rat and human) and a perfused-pig-ear model. Percutaneous penetration of propoxur in human volunteers was measured by analysis of its metabolite (2-isopropoxyphenol) in blood and urine; in all other studies radiolabeled propoxur ([ring-U-(14)C]propoxur) was used. In order to allow for direct comparison, experimental conditions were standardized with respect to dose (150 microg propoxur per cm(2)), vehicle (60% aqueous ethanol) and exposure time (4 h). In human volunteers, it was found that approximately 6% of the applied dose was excreted via the urine after 24 h, while the potential absorbed dose (amount applied minus amount washed off) was 23 microg/cm(2). In rats these values were 21% and 88 microg/cm(2), respectively. Data obtained in vitro were almost always higher than those obtained in human volunteers. The most accurate in vitro prediction of the human in vivo percutaneous absorption of propoxur was obtained on the basis of the potential absorbed dose. The absorbed dose and the maximal flux in viable full-thickness skin membranes correlated reasonably well with the human in vivo situation (maximal overestimation by a factor of 3). Epidermal membranes overestimated the human in vivo data up to a factor of 8, but the species-differences observed in vivo were reflected correctly in this model. The data generated in the perfused-pig-ear model were generally intermediate between viable skin membranes and epidermal membranes.