Objective: To assess the outcomes in women with epilepsy in relation to fetal malformation and epileptic seizure control during pregnancy when valproate (VPA) intake was ceased, or the drug's dose was reduced before pregnancy. Methods: Statistical analysis of data collected in the Australian Pregnancy Register between 1999 and 2018 concerning 580 pregnancies previously treated with VPA, with the VPA dose reduced or the drug withdrawn prior to pregnancy in 158 cases. Results: Although the available data have limitations, fetal malformation rates in the pregnancies studied were lower in the VPA changed pregnancies (4.5%) than in the VPA unchanged comparator pregnancies (10.9%, hazard ratio [HR] 0.412, 95% confidence interval [CI] 0.190-0.892), and were only 2.7% where VPA intake was ceased before pregnancy (HR 0.262, 95% CI 0.083-0.826). Seizure-affected pregnancies were more frequent in the VPA changed pregnancies than in the VPA unchanged ones (46.2% vs 30.8%, HR 1.500, 95% CI 1.203-1.870). Convulsive seizure-affected pregnancies also were increased, but the difference was not statistically significant. Significance: Prepregnancy reduction in VPA dosage reduced the hazard of fetal malformations, whereas ceasing intake of the drug decreased the hazard to one similar to that which applies in the general population, but at a cost of decreased control of epileptic seizures during the pregnancies studied. Further investigations are needed to see whether such findings apply more widely in women with epilepsy.