In current cutaneous retinoid therapy systemic exposure is low and the risk of teratogenesis appears to be limited. However, new indications, altered posologies and the introduction of new synthetic retinoids demand continuous assessment of the teratogenic risk. Teratogenicity testing of new substances in animals is only of value if accompanied by detailed pharmacokinetic analysis to establish the relationships between the levels of parent compound and metabolite in both maternal plasma and fetal tissues and teratogenic events. This information should be compared to the maximum of relevant pharmacokinetic data which can be ethically obtained in man or from human tissues. The presence or absence of teratogenic effects following cutaneous administration of retinoids in animals has, as such, little direct bearing on the risk in man. Two special cases exist where teratogenic risk can be evaluated directly in man without reference to animal studies. The first concerns substances whose teratogenic potential has been established in man by other routes of administration permitting a comparison with the cutaneous route on a pharmacokinetic basis. The second concerns the cutaneous administration of endogenously occurring substances and their eventual disturbance of systemic retinoid equilibrium.