Abstract: Intratumoral heterogeneity is a key feature of solid tumors, including clonal heterogeneity, spatial heterogeneity and temporal heterogeneity. It plays a significant role in tumor growth, proliferation, invasion, metastasis, and treatment sensitivity, which is closely associated with adverse prognoses in cancer patients.Despite significant progress in the assessment of intratumor heterogeneity in recent years, effective means for early and non‑invasive assessment of intratumor heterogeneity are still lacking. Radiomics, leveraging computer technology to extract features from tumor images, provides quantitative analysis with the advantages of non‑invasiveness and systematic analysis. In recent years, radiomics has demonstrated great potential in the early diagnosis, therapeutic response prediction, and prognosis assessment of intratumor heterogeneity. By integrating multiple machine learning and deep learning modeling analyses, it can provide important references and support for clinical decision‑making related to tumor heterogeneity. Furthermore, combining radiomics with other omics can provide a new perspective for analyzing the connection between macroscopic phenomena and microscopic biology.The authors review the progress and clinical application potential of radiomics in the assessment of intratumoral heterogeneity in recent years, with the aim of providing some reference and reflection for radiomics‑driven precise diagnosis and treatment of tumors.