Advances in the molecular characterization of leukemic cells have greatly improved the precision of diagnosis and treatment assignment as well as the monitoring of residual disease in both acute lymphoblastic leukemia and acute myeloid leukemia. Currently, specific genetic rearrangements can be identified in as many as 50% of children with either acute lymphoblastic leukemia or acute myeloid leukemia. The genes p16 (or MTS1) and TEL/AML1 are now respectively recognized as the most common tumor suppressor and fusion genes in childhood acute lymphoblastic leukemia. Increasingly, contemporary protocols for the acute leukemias are relying on genetic information to guide treatment decisions. Examples include the use of allogeneic hematopoietic stem cell transplantation for acute lymphoblastic leukemia with the BCR-ABL fusion gene or MLL rearrangement, and for acute myeloid leukemia with monosomy 7; antimetabolite-based therapy for acute lymphoblastic leukemia cases with hyperdiploidy of more than 50 chromosomes (DNA index > or = 1.16); and retinoic acid and anthracycline-containing regimens for the acute promyelocytic acute myeloid leukemia subtype with PML-RARA fusion. Other efforts are being made to reduce the long-term sequelae of treatment. Indeed, extended intrathecal therapy and intensive systemic chemotherapy will, in all likelihood, replace cranial irradiation as subclinical central nervous system therapy for patients with intermediate-risk acute lymphoblastic leukemia, and perhaps even for those with high-risk acute lymphoblastic leukemia. The challenge now is to identify specific treatments for other genetically defined subtypes of leukemia. This goal will be realized only through protocol-based studies employing uniform criteria for defining risk status.