A still unresolved public health concern is that excessive vitamin A intake, like vitamin A deficiency, possibly causes birth defects not only in animals but also in man. Due to the low incidence of possibly vitamin A-related malformations in man, available data cannot convincingly define the upper safe limit of periconceptional vitamin A intake. Direct human intervention studies are not feasible for ethical reasons. Therefore, a novel approach in addressing this issue was chosen by combining teratogenicity data from a validated animal model with data on systemic exposure to vitamin A and its major metabolites in female volunteers. In a study in pregnant women endogenous plasma concentrations of vitamin A metabolites during early pregnancy ranged from 0.26 to 7.72 ng/ml. Since they did not cause any foetal malformations, retinoid plasma levels in this range can be considered non-teratogenic. Results of a trial in non-pregnant women document that daily oral vitamin A supplements of 4000, 10,000 and 30,000 IU given for 3 weeks were in the range or slightly above the range of endogenous plasma levels seen in early pregnancy. Even after a 3-week treatment with 30,000 IU/day, peak plasma levels of retinoic acid and isotretinoin were within or just slightly above the range of their physiological levels. In cynomolgus monkeys (average weight: 3-4 kg), a NOAEL (no observed adverse effect level) of 7500 IU per kg body weight and a LOAEL (lowest observed adverse effect level) for developmental toxicity of 20,000 IU/kg was found. Considering these results in the cynomolgus monkey, a dose of 30,000 IU/day should also be considered as non-teratogenic in man.