Epidemiological hypotheses on disease etiology, generated by the observation of geographic distribution and time trends, can be confirmed or refuted by analytical investigations on specific risk factors. In the case of leukemias, lymphomas and myelomas, however, hypothesis generation is limited by the use of the ICD classification in mortality and incidence statistics. We compared recent incidence data in different parts of the world and at different times for leukemias, lymphomas and myelomas. The incidence rate of non-Hodgkin's lymphomas (NHL) is increasing in most Western countries, while trends for the other hematolymphopoietic malignancies are strikingly stable. To formulate hypotheses on the causes of this pattern would require a more appropriate classification of descriptive data. Excesses of non-Hodgkin's lymphomas have been observed in populations exposed to phenoxy-acetic acid herbicides, to insecticides and to organic solvents. Some of these exposures, in particular TCDD, which is a contaminant of phenoxy herbicides, DDT and chlorinated solvents, have been reported to alter cell-mediated immunity. The incidence of NHL is also increased among subjects with HIV infection and subjects undergoing heart or kidney transplantation, all of whom experience immunodeficiency. A hypothesis that has been put forward recently is that the NHL increase is related to increased exposure to sunlight, which has immunosuppressive effects. From a mechanistic point of view, one can hypothesize that NHL is caused by exposures that induce proliferation and immortalization of B-cells, followed by T-cell impairment entailing cell-mediated immune deficiency.