Abstract Major Depression (MD) is a prevalent, disabling and life-limiting condition. The neurobiological associations of genetic risk for MD remain under-explored in large samples, with no comprehensive mega-analysis conducted to date. Our study analysed data from 11 separate studies, encompassing 50,975 participants from the ENIGMA Major Depressive Disorder Working Group. We developed highly consistent genetic and neuroimaging protocols and applied these throughout all participating studies, together with rigorous genetic methods to remove overlap between the polygenic risk scores (PRS) training and testing samples. Elevated PRS for MD correlated with lower intracranial volume and lower global measure of cortical surface area (βICV = −0.017, pICV = 1.97 × 10−6; βSurf = −0.013, pSurf = 4.5 × 10−4; pFDR < 3.62 × 10−4). The most significant cortical association was observed in the surface area of the frontal lobe (β = −0.011, p = 2.85 × 10−6, pFDR = 1.42 × 10−5), particularly in the left medial orbito-frontal gyrus (β = −0.021, p = 9.48 × 10−8, pFDR = 1.25 × 10−5). In subcortical regions, lower volumes of the thalamus, hippocampus, and pallidum correlated with higher PRS of MD (β ranged from −0.011 to −0.015, p ranged from 0.002–1.73 × 10−5, pFDR < 0.006). In a subsample of young individuals only (<25 years old, N = 5570), although there were no FDR-significant findings, directions of effects were highly consistent between the analyses of cortical surface areas in youth and the full sample (71.2% in the same direction, exact binomial test p-value = 7.56 × 10−4). Subsequent Mendelian randomisation analysis revealed potentially causal effects of smaller left hippocampal volume on higher liability for MD (Inverse variance weighted analysis β = −0.064, p = 8.04 × 10−3, pFDR = 0.04). Our findings represent an example of how extensive international collaborations can significantly advance our neurogenetic understanding of MD and give insights to avenues for early interventions in those at high risk for developing MD.