Abstract: Intrahepatic cholangiocarcinoma (ICC) has a high rate of lymph node metastasis, which is an independent risk factor affecting prognosis. However, the necessity of routine lymph node dissection (LND) remain controversial. Several clinical guidelines recommend regional LND should be considered post‑surgery to obtain an accurate N staging for ICC. Nonetheless, the specific extent, number of lymph nodes, and optimal timing of LND remain unresolved. Although routine LND facilitates accurate N staging, its clinical value in improving long‑term prognosis is uncertain. Furthermore, lymph nodes play a role in tumor immunoregulation, and unnecessary dissection may compromise immune function. In response to these dilemmas, a strategy named selective lymph node dissection has been proposed, aiming to identify patients most likely to benefit based on clinical and biological profiles: sparing low‑risk patients while ensuring high‑risk individuals receive standardized dissection. However, this strategy still face four challenges, such as high rates of undetected micrometastases on preoperative imaging, limitations of intraoperative frozen section analysis, poorly defined lymphatic drainage patterns, and insufficient evidence‑based guidance. To address these, the authors systematically examine the paradigm shift from imaging anatomy to biological behavior in guiding LND for ICC. Key advances include radiomics and molecular markers for risk stratification, tracer‑guided navigation for precise localization, liquid biopsy for early detection, and artificial intelligence integrated with multi‑omics for biological assessment.