To evaluate corpus callosum (CC) morphometry by MRI in term small for gestational age (SGA) fetuses compared to controls and its association with neurobehavioral outcome. 117 SGA and 73 control fetuses with normal umbilical Doppler were imaged using a 3T MRI scanner at 37 weeks of gestation, obtaining T2 HASTE anatomical acquisition in the three orthogonal planes. CC length, thickness, total area and the areas after a subdivision in 7 portions were blindly assessed by semi-automatic delineation. All measurements were corrected by the cephalic index. Neonatal neurobehaviour was evaluated with the Neonatal Behavioral Assessment Scale (NBAS) at 42 ± 1 week. Statistical analyses were performed by multivariate analysis of covariance adjusting for gender, smoking and body mass index. The association between CC measurements and NBAS scores in SGA was assessed by ordinal regression taking into account the same covariates plus age at NBAS test, considering the dependent variable the amount of NBAS test's clusters under the 5th centile. SGA fetuses showed significantly smaller total CC area (SGA: 1.3996 ± 0.26 vs. control: 1.664 ± 0.31; p < 0.01) and smaller genu, rostral body, anterior midbody, posterior midbody, isthmus and splenium areas. Overall, we found a significant association between smaller CC measures and worse neurobehavioral outcome in SGA newborns, particularly a smaller total area (p = 0.03), rostrum (p = 0.02), genu (p = 0.03), rostral body (p = 0.02), anterior midbody (p = 0.03) and splenium (p = 0.03) had a significant negative effect on the number of abnormal NBAS clusters. CC development was significantly altered in term SGA fetuses and correlated with worse neurobehavioral performance. CC could be further explored as a potential imaging biomarker to predict abnormal neurodevelopment in pregnancies at risk.