Abstract Aim: To identify easily accessible neuroanatomical abnormalities useful for diagnosing fetal alcohol spectrum disorders (FASD) in fetal alcohol syndrome (FAS) but more importantly for the probabilistic diagnosis of non-­syndromic forms (NS-­FASD). Method: We retrospectively collected monocentric data from 52 individuals with FAS, 37 with NS-­FASD, and 94 paired typically developing individuals (6–­20 years, 99 males, 84 females). On brain T1-­weighted magnetic resonance imaging, we measured brain size, corpus callosum length and thicknesses, vermis height, then evaluated vermis foliation (Likert scale). For each parameter, we established variations with age and brain size in comparison individuals (growth and scaling charts), then identified participants with abnormal measurements (<10th centile). Results: According to growth charts, there was an excess of FAS with abnormally small brain, isthmus, splenium, and vermis. According to scaling charts, this excess remained only for isthmus thickness and vermis height. The vermis foliation was pathological in 18% of those with FASD but in no comparison individual. Overall, 39% of those with FAS, 27% with NS-­FASD, but only 2% of comparison individuals presented with two FAS-­recurrent abnormalities, and 19% of those with FAS had all three. Considering the number of anomalies, there was a higher likelihood of a causal link with alcohol in 14% of those with NS-­FASD. Interpretation: Our results suggest that adding an explicit composite neuroanatomical–­radiological criterion for FASD diagnosis may improve its specificity, especially in NS-­FASD.