Existing systems of staging for patients with rectal cancer depend almost exclusively on anatomic evidence. Consequently, the stages cannot be determined in advance of therapeutic decisions and cannot be used for patients treated without surgery. Furthermore, the stages contain no provision for important prognostic distinctions, that cannot be discerned from anatomic data. After preparing a taxonomy for hiterto unclassified medical data, we developed and tested two new systems of staging in a cohort if 318 patients. The first system which can be applied before treatment, is divided into four composite stages that contain elements of symptomatic, chronometric, co-morbid, and para-morbid data, as well as information obtained from physical examination, sigmoidoscopy, and roentgenography. The second system, applicable to patients with resected tumors, is based on a combination of pre-therapeutic clinical information and post-surgical anatomic evidence. The two systems produce prognostic gradients that are clinically distinctive and statistically efficacious.