Cortisone administered during pregnancy prevents the phenotypic expression of lidgap-Miller (lgM1), a genetically determined congenital defect in mice. The dose-response, on a probit log dose scale, is linear to a dose of 60 mg/kg maternal body weight, giving an observed maximum cure of 94% and an ED50(1) of 27 mg/kg. At higher doses the rate of prevention diminishes, to 56% at 320 mg/kg. Fetal (18-day) weight remains unaffected for doses of up to 60 mg/kg and then drops sharply at higher doses, suggesting that cortisone is toxic in this higher range. The severity of lidgap, measured in two ways (as one or both eyes open and as width of gap), decreases as its frequency falls. This finding is in keeping with the threshold model for birth defects. The threshold appears to be in the embryos rather than in their mothers. There is a broad optimum time of treatment: days 13-14 for the 80 mg/kg dose, and days 14-15 for the 30 mg/kg dose. Treatment on day 16 has little or no effect. It is not known how cortisone cures the lidgap trait. The action may be specific--to stimulate transcription of a specific gene, possibly of the lidgap locus itself--or more general, e.g., to alter cell division rate or extracellular matrix constituents.