Introduction: The incidence of squamous cell carcinoma of the anus (SCCA) has increased 3- to 4-fold in western populations over the past 30 years. Initial treatment is chemoradiotherapy (CRT) but is associated with treatment-related morbidity. Up to 30% of patients may suffer locoregional failure (LRF) that requires radical salvage surgery, which in turn is associated with further morbidity. Currently, all patients with SCCA are broadly treated with the same CRT regimen. There is a need to balance reduced treatment-related morbidity while reducing LRF rates through personalised approaches. Over the past decade, modern imaging modalities, notably MRI, have been introduced into UK clinical practice to better personalise CRT but has not been evaluated at large-scale. Here, this was addressed in five inter-related chapters. Chapters 3 & 4: Nodal disease (LN+) is traditionally considered an adverse prognosticator, but its significance in an era of image-enhanced detection is uncertain. Through systematic review (62 datasets) and detailed analysis of The Christie SCCA cohort (1990 to 2014), I demonstrated a substantial image-driven upward nodal stage-migration, associated with improved survival in both LN+ and LN- patients (The Will Rogers phenomenon) and added a novel finding, which I termed reduced prognostic discrimination. This was published in Lancet Oncology. Chapters 5 & 6: I undertook a MRI evaluation of tumour and nodal characteristics in 265 patients with SCCA treated by CRT at The Christie. With in-house radiologists, I developed a protocol for the detection of tumour and nodal features and validated this both internally (two Christie radiologists) and externally (one Leeds radiologist). As expected, tumour size (as assessed by mrT-stage and mrT-size) was a key prognostic factor for overall survival and LRF. I identified novel prognostic factors ÃƒÂ¢Ã‚Â€Ã‚Â“ namely extra-mural vascular invasion and tumour signal heterogeneity. I demonstrated that pre-sacral nodal involvement might represent a particularly poor prognostic perirectal nodal field sub-group. In a sub-study of 81 patients from The Christie and Leeds (40 with LRF; 41 LRF-free), I measured tumour volume using 3D contouring and found that volume is a potentially superior discriminator for LRF over traditional T-size parameters. Chapter 7: In collaboration with the Radiotherapy Related Research Group, I performed image registration techniques to examine patterns of LRF in 269 patients; I demonstrated that a third of LRFs following CRT involved nodal failures, with different mechanisms of pelvic failure present. Primary anorectal failures and the majority of mesorectal and inguinal nodal failures occurred following (adequate) high-dose radiotherapy (RT); while pre-sacral and iliac nodal failures occurred above the superior RT field. I constructed average 3D dose-distribution maps and found no evidence for incidental dose variations being associated with LRF. Discussion: The findings in Chapters 3 & 4 serve as a reminder that the introduction of new staging technologies in oncology might misclassify true disease stage, spuriously informing disease management and ultimately increasing the risk of overtreatment. The findings in Chapters 5 & 6 offer novel imaging biomarkers, with (radio-)biological plausibility, to be evaluated in future studies to optimise treatments and reduce LRF. Chapter 7 serves two key purposes: a reminder that the anorectum is the key failure site, despite adequate RT doses; and the dose maps will be a platform to benchmark new IMRT approaches and disease control.