The percutaneous absorption of retinol (Vitamin A) from cosmetic formulations was studied to predict systemic absorption and to understand the significance of the skin reservoir in in vitro absorption studies. Viable skin from fuzzy rat or human subjects was assembled in flow-through diffusion cells for in vitro absorption studies. In vivo absorption studies using fuzzy rats were performed in glass metabolism cages for collection of urine, feces, and body content. Retinol (0.3%) formulations (hydroalcoholic gel and oil-in-water emulsion) containing (3)H-retinol were applied and absorption was measured at 24 or 72 h. All percentages reported are % of applied dose. In vitro studies using human skin and the gel and emulsion vehicles found 0.3 and 1.3% retinol, respectively, in receptor fluid at 24 h. Levels of absorption in the receptor fluid increased over 72 h with the gel and emulsion vehicles. Using the gel vehicle, in vitro rat skin studies found 23% in skin and 6% in receptor fluid at 24 h, while 72-h studies found 18% in skin and 13% in receptor fluid. Thus, significant amounts of retinol remained in rat skin at 24 h and decreased over 72 h, with proportional increases in receptor fluid. In vivo rat studies with the gel found 4% systemic absorption of retinol after 24 h and systemic absorption did not increase at 72 h. Retinol remaining in rat skin after in vivo application was 18% and 13% of the applied dermal dose after 24 and 72 h, respectively. Similar observations were made with the oil-in water emulsion vehicle in the rat. Retinol formed a reservoir in rat skin both in vivo and in vitro. Little additional retinol was bioavailable after 24 h. Comparison of these in vitro and in vivo results for absorption through rat skin indicates that the 24-h in vitro receptor fluid value accurately estimated 24-h in vivo systemic absorption. Therefore, the best single estimate of retinol systemic absorption from in vitro human skin studies is the 24-h receptor fluid value. However, the receptor fluid value from the 72-h extended study may be used in a worst-case exposure estimate. In conclusion, in vivo skin absorption studies can be useful in determining whether to include material in the in vitro skin reservoir as absorbable material in estimates of systemic absorption.