How to accurately identify factors of cancer occurrence and to provide intervention early are the key issues that urgently need to be addressed in cancer prevention and treatment. Mendelian randomization (MR) analysis uses genetic variants as instrument variables for exposures of interest, which compensates the shortcomings of traditional observational studies and clinical trials. This review introduced the current application status of MR analysis in cancer etiology and treatment researches in details, including assessment of cancer risk factors, exploration of cancer treatment targets, and evaluation of drug efficiency and adverse reactions. The scopes and dimensions of cancer etiology and treatment researches are greatly expanded because of various MR designs and abundant high-level omics data. As well, it provides a practical and feasible method for constructing cancer etiology networks and drug targeted databases, which are promising for supporting the development of precision cancer prevention and treatment.