Comprehensive genomic testing in routine cancer care pathways has created the need to interpret the consequences of somatic (acquired) genomic variants beyond the currently well-characterised driver variants in cancer gene hotspots. While several guidelines have been published to determine the oncogenicity of somatic cancer gene variants, they lack a comprehensive and flexible approach that encompasses all available lines of evidence. Individual UK laboratories have developed local approaches to standardise somatic variant interpretation, often based on different sets of published guidelines, but a comprehensive national standardised framework is lacking. The absence of standardisation in approaches to somatic variant interpretation highlights a significant gap in the field of genomic medicine within the UK healthcare system. Key stakeholders from across the UK cancer genomics diagnostic community formed the UK Somatic Variant Interpretation Group (SVIG-UK) in September 2018 to develop a consensus approach for interpretation of somatic variants identified through genomic testing in patients with solid tumours and haematological malignancies. SVIG-UK scientists conducted a review of existing somatic variant interpretation classification systems and although they mostly agreed on evidence sources for variant interpretation, differences were identified in how the evidence should be used, weighted and combined. The SVIG-UK team subsequently developed a single, standardised UK-wide approach to somatic variant interpretation which encompassed both solid tumour and haematological cancer genomic testing. This framework was shared with stakeholders across the UK alongside variants for preliminary testing. Outcomes were then reviewed and following engagement sessions across the community, the variant interpretation recommendations were updated and ratified by the UK Association of Clinical Genomics Sciences. We present herein the SVIG-UK framework and recommendations, which provide a standardised, comprehensive and flexible approach for classifying the oncogenicity of somatic variants in cancer genes.