The correct and efficient repair of DNA Double-Strand Breaks plays an important role in both preventing the development of cancer, but also in the ability of cancer cells to resist ionising radiation. Cancer associated mutations have been identified throughout all the components of the non homologous end joining (NHEJ) pathway, one of the main methods to repair DSBs. We aimed to use information on how these mutations affected the protein structure and function in order to better understand the underlying molecular mechanisms, and to see if we could use this to better identify those mutations most likely to be cancer drivers. Over 1,131 cancer associated variants and 1,852 population variants were mapped to the principle components of the NHEJ system- DNA-Lig IV, DNA-PKcs, Ku70/80 and XRCC4. The molecular effects of each variant was assessed using the mCSM platform to quantitatively assess their impact on protein folding, dynamics, stability and interactions. We observed that cancer driving variants in the scaffolding proteins, such as XRCC4, were more likely to affect key protein-protein interactions, whilst those in the enzymatic components, such as DNA-LigIV and DNA-PKcs, the effects on interactions with binding components were key factors. Using this insight, we built predictive classifiers that were able to accurately identify driving mutations, up to 80% accuracy, 72% precision, 0.78 AUC. This highlights the power of using protein structural information to assess the molecular consequences of genomic variants and better understand their link to phenotypic outcomes.