An essential purpose of a pathologic classification of non-Hodgkin's lymphomas is to supply guidance in the clinical management of patients. Ideally, an optimal subclassification should also be scientifically accurate, highly reproducible, and readily teachable. Such a system, when used in conjunction with uniform staging, should enable relatively homogeneous groups of patients to be defined. In the present study, we have evaluated four new systems for possible additions to the traditional Rappaport classifcation. In response to specific questions the following tentative conclusions could be drawn: (1) Within the Rappaport nodular lymphomas, there are no differences in survival between lymphomas that are totally nodular verusus those that are nodular and diffuse. (2) In lymphomas composed of small cleaved follicular center cells of the Lukes-Collins system, survival appears to be independent of pattern (follicular, follicular and diffuse, or diffuse). In contrast, in tumors classified as centroblastic-centrocytic in the Kiel classification or those classified as large cleaved or large noncleaved in the Lukes-Collins system, a totally or partially follicular pattern confers a better prognosis than its diffuse counterpart. (3) The numbers are small but there is no apparent difference in survival between cases of Rappaport's difuse well differentiated lymphocytic lymphomas with or without plasmacytoid differentiation. (4) Within the original Rappaport DPDL there were at least two distinct types of lymphomas: (1) a convoluted lymphoblastic that occurs in younger patients has a high frequency of B symptoms and carries a poor prognosis; and (2) a diffuse lymphoma that is cytologically identical to nodular PDL, occurs in older patients, and has a relatively good prognosis. In August of 1976 Rappaport modified his classification to recognize these lymphoblastic lymphomas as a distinct clinicopathologic entity. (5) In this 22-yr retrospective review, neither the Kiel nor the Lukes-Collins system could identify any relatively favorable subsets within Rappaport's category of diffuse histiocytic lymphoma. Prospective studies applying the same approach to large numbers of patients subjected to modern uniform staging and aggressive combination chemotherapy may provide data upon which to base an optimal subclassification of DHL.