The epidemiologic features of acute leukemia in children and adults are reviewed and suggestions are made for new research directions. Similarities in risk factors exist for adults and children, but differences in tumor biology suggest that the etiological factors may not be the same. Many unifying hypotheses have been proposed for childhood leukemia based on well-designed studies. Two recent theories, both involving an unusual response to childhood infection, attempt to explain the 2- to 5-year age peak. The etiology of infant leukemia may be related to chromosomal damage sustained in utero, analogous to the damage in adults with treatment-related leukemia. Although much is known about risk for treatment-related leukemias in adults, environmental, occupational, and other risk factors are less well studied. More homogeneous populations are needed for study including populations where distinction is made between acute myleloid leukemia patients with and without antecedent hematologic disorders.