Many plate-based in vitro assays of membrane permeability (e.g., Caco-2, MDCK, PAMPA) of sparingly soluble candidate molecules report permeability of water, and not of the intended membrane barrier. This is so because the unstirred water layer on both sides of the membrane barrier is rate limiting for these highly permeable molecules. The thickness of this water layer can be 1500-4000 microm in unstirred assays. Under in vivo conditions, however, the unstirred water layer is believed to be 30-100 microm thick. Lightly stirred in vitro assays, using plate shakers, cannot lower the thickness of the water layer to match that found in vivo. In this study, 55 lipophilic drugs were employed to characterize the effect of stirring in parallel artificial membrane permeability assay (PAMPA). Highly efficient individual-well magnetic stirring at speeds greater than 110 rpm has been demonstrated to lower the unstirred water layer thickness to the in vivo range. Stirring at 622 rpm has lowered the layer thickness to 13 microm in some cases, which had not been previously achieved for plate-based permeability assays. With diminished water layer contribution at 622 rpm, for example, the effective permeability of progesterone is 2754 x 10(-6) cm/s. The new stirring apparatus used in this study is not only suitable for PAMPA, but can also be used in Caco-2 assays. Because of the diminished resistance of the thinner water layer, the stirred PAMPA permeation time has decreased from the usual 15 h to about 15 min for lipophilic compounds.