There is little evidence that time-space clustering is important in the epidemiology of leukemia, lymphoma, Burkitt's tumor, and myeloma. Clustering by interpersonal contact, particularly in Hodgkin's disease, does appear to occur, but more study is needed to warrant the conclusion that such contact is necessary for subsequent illness to develop. Future epidemiologic studies should include the use of laboratory analysis to evaluate the role of viruses, genetics, and environmental factors in cluster studies.