Selenium (Se)-dependent and -independent glutathione (GSH) peroxidases detoxify H2O2 and lipid hydroperoxides, which may mediate the teratogenicity of phenytoin and related xenobiotics. To test this hypothesis, CD-1 mice were placed on Se-deficient diets for 15, 25 or 40 days and bred so that the day of analysis corresponded to gestational day 11. In Se-replete control animals, embryonic peroxidase activities were only 5% of activities in maternal liver (P < .05). After 15 days of Se deprivation, maternal activities for H2O2 (reflecting Se-dependent peroxidase) and cumene hydroperoxide (CmOOH) (reflecting both Se-dependent and -independent peroxidases) were reduced to 20% (P < .05) and 35% of controls, respectively. At this time, the incidence of fetal cleft palates initiated by phenytoin (55 mg/kg intraperitoneally on gestational days 11, 12 and 13) was doubled, from 12% to 25% (P < .05). Selenite rescue (Na2SeO3, 350 micrograms/kg intraperitoneally on day 9) restored maternal and embryonic peroxidase activities and completely inhibited phenytoin-initiated postpartum lethality and fetal resorptions in animals that had been Se depleted for 15 days. After 40 days of Se deprivation, maternal and embryonic peroxidase/H2O2 activities were reduced to < 1% and 27% of Se-replete controls, respectively. In contrast, maternal peroxidase/CmOOH activity was increased to 70% of controls, reflecting induction of Se-independent peroxidase, compared with that with 15 days' depletion. Phenytoin-initiated cleft palates with 40 days' depletion appeared to be reduced (16%) compared with Se-replete controls (24%) (P < .07). In 40-day Se-depleted animals given selenite rescue, the 10% incidence of cleft palates was significantly lower than that in the 40-day Se-replete group (24%) but not the Se-depleted group (16%). This is the first demonstration of reduced Se-dependent GSH peroxidase activities in embryonic tissues with dietary Se-deprivation. The results implicate reactive oxygen species and lipid hydroperoxides in the mechanism of phenytoin teratogenicity and suggest that GSH peroxidases are important embryoprotective enzymes.