Pathologic (Dukes) and clinicopathologic staging systems (Australian and TNM) are all used for the staging of colorectal cancer. Many modifications of the systems, with different standards of evidence, are currently used for a variety of purposes. It is not yet possible to readily exchange data from one staging system to another because of the lack of uniform anatomic and pathologic subdivisions. It would also be an advantage if staging systems more confidently predicted potential for cure. It is proposed that the TNM based system be modified to satisfy these requirements. Modifications identify common ground between the various systems, use prognostic variables shown to be significant by multivariate analyses, and introduce an abbreviated form of tumor grading. Grading improves prediction of cure and allows standardization of the composition of stages and substages because distant metastatic potential is shown to be related to the presence of high grade. The modifications are considered necessary and appropriate for epidemiologic and clinical studies.