Radical radiotherapy for prostate cancer offers excellent long-term outcomes for patients with high-risk disease. The increased risk of pelvic nodal involvement in this cohort has led to the development of whole-pelvis radiotherapy (WPRT) with a prostate boost. However, the use of WPRT remains controversial. Data are mixed, but advanced radiotherapy techniques enable delivery of increased radiation to pelvic nodes with acceptable levels of toxicity. Contemporary imaging modalities with increased sensitivity for detecting subclinical lymph node disease will facilitate selection of patients most likely to benefit from WPRT. Using such modalities for image guidance of advanced radiotherapy techniques could also permit high-dose delivery to nodes outside the conventional Radiation Therapy Oncology Group volumes, where magnetic resonance lymphography and single-photon-emission CT imaging have mapped a high frequency of microscopic disease. With increased toxicity a concern, an alternative to WPRT would be selective irradiation of target nodal groups most likely to harbour occult disease. New image-based ‘big data’ mining techniques enable the large-scale comparison of incidental dose distributions of thousands of patients treated in the past. By using novel computing methods and artificial intelligence, high-risk regions can be identified and used to optimize WPRT through refined knowledge of the likely location of subclinical disease.