Effects of valproate (VPA) dose and treatment discontinuation during the first trimester of pregnancy on the risks of spontaneous abortions (SAB) and major birth defects were analyzed. Pregnancies with first trimester VPA exposure (n = 484) prospectively recorded by the German Embryotox center in 1997–2016 were compared with a randomly selected, non-exposed cohort (n = 1446). The SAB risk was not significantly increased in the VPA cohort [HRadj 1.31 (95% CI 0.85–2.02)] but major birth defects were significantly more frequent [8.7% vs. 3.4%; ORadj 2.61 (95% CI 1.51–4.50)]. Risk was even higher in pregnancies with no VPA discontinuation in first trimester [ORadj 3.66 (95% CI 2.04–6.54)]. Significant ORs were found for nervous system defects in general [ORadj 5.69 (95% CI 1.73–18.78)], severe microcephaly [ORadj 6.65 (95% CI 1.17–37.68)], hypospadias [ORadj 19.49 (95% CI 1.80–211)] and urinary system defects [ORadj 6.51 (95% CI 1.48–28.67)]. VPA dose had a stronger effect than antiepileptic poly- versus monotherapy; for VPA dose ≥ 1500 mg/day the ORadj was 5.41 (95% CI 2.32–12.66)]. A daily dose increase of 100 mg was calculated to raise the risk for major birth defects by 15% [OR 1.15 (95% CI 1.08–1.23)]. Overall, maternal first trimester treatment regimen had a relevant impact on birth defect risk.