Abstract： Cancer is one of the major causes of human death. Early diagnosis and treatment are widely believed to be important in decreasing cancer mortality rate. However, biological factors (such as tumor heterogeneity) and social factors (such as profits driving), over-diagnosis and over-treatment exist in the practice of cancer diagnosis and treatment.The keys to avoid such issues are to identify what kinds of precancerous lesions are more likely to progress into malignant tumors,what kind of early cancer are prone to invading surrounding tissues, and what kinds of prophylactic and therapeutic options are of benefit to effective survival of patients.All above need to be accurately evaluated by cohort studies.Cohort study is the most reliable method to crystallize causal relationship of the exposures(interventions)with the development and recurrence of cancers and define the directions of cancer screening and therapy. Cohort studies shed light on the prediction,screening,and active prophylaxis for the occurrence and recurrence of cancers with specific types. Moreover, the prognostic effects of a given clinical treatment can be precisely evaluated in cohort study, rather than experience-directed clinical activity. Therefore, cohort study is indispensable for cancer research.