The human skin grafted congenitally athymic (nude) mouse, pig skin grafted nude mouse, hairless dog, and weanling Yorkshire pig were evaluated as models for predicting skin penetration in man. Nine radiolabelled compounds previously tested on man were applied topically (4 micrograms/cm2) to each model. These compounds included caffeine, benzoic acid, N, N-diethyl-m-toluamide, three steroids, and three insecticides. To correct for incomplete excretion of the label following topical absorption, per cent penetration was calculated by dividing the per cent of the topically applied radioactive dose recovered in the excreta by the corresponding percentage after parenteral administration and multiplication by 100. Calculated values of per cent penetration were confirmed in the case of the grafted nude mouse because significant correlations (r = 0.78 for human skin grafted athymic nude mouse and r = 0.97 for pig skin grafted athymic nude mouse) were found between the calculated values and the actual values obtained by summing the radioactivity recovered in the urine, faeces, tissues, and carcass. The results also revealed a significant correlation between human skin grafted athymic nude mouse values and human values (r = 0.74, P = 0.05) and between weanling Yorkshire pig values and human values (r = 0.83, P = 0.05). In contrast, no significant correlation existed between human values and those of the hairless dog and the pig skin grafted athymic nude mouse.