Abstract The in vivo teratogenic potential of valproic acid (VPA) and related teratogenic and non-teratogenic analogues has been correlated with their effects on specific in vitro endpoints of cell proliferation, migration and CAM-dependent neurite outgrowth, as these events are common to crucial epochs of development. The (±)-2- n-propyl-4-pentynoic acid [(±)-4-yn-VPA] and S-2- n-propyl-4-pentynoic acid [ S(-)-4-yn-VPA] analogues increased the incidence of neural tube defects in mouse embryos exposed to a single dose, whereas the E-2- n-propyl-2-pentenoic acid ( E-2-en-VPA) analogue and R-2- n-propyl-4-pentynoic acid [ R( + )-4-yn-VPA] enantiomer were without effect. VPA and related analogues tested exerted comparable G1 phase antiproliferative effects in C6 glioma and limb bud cells in a dose range of 0–3 mM; however, their relative potency did not correlate with in vivo teratogenicity. In contrast, VPA and all teratogenic analogues, at 3 mM, inhibited neuronal cell aggregation and limb bud chondrocyte differentiation in a manner that exhibited a reasonable correlation with their in vivo teratogenicity. The teratogenic S(−)-4-yn-VPA and non-teratogenic R( + )-4-yn-VPA enantiomers exhibited a differential inhibition of primary neurone outgrowth of neuntes stimulated by cell adhesion molecules [L1 and N-cadherin (NCAD)]. Half-maximal inhibition was observed at approximately 150 μM for the teratogenic S(−)-4-yn-VPA enantiomer, but not the non-teratogenic R( + )-4-yn-VPA form. These results suggest that in vitro perturbations of differentiation are likely to provide the greatest discriminatory power for in vivo teratogenicity.