The aim of this study is to clarify the clinicopathologic characteristics of adolescent leukemia in Japan by retrospective analysis. Patients with acute lymphoblastic leukemia (ALL), acute myelogenous leukemia (AML), chronic myelogenous leukemia (CML), and myelodysplastic syndrome (MDS), consecutively diagnosed from 1986 to 1999, were enrolled. A total of 3,856 patients from 1 to 15 years of age and 1,803 patients from 15 to 29 years of age were eligible for this study. Demographically, the frequency of AML found was almost constant during the teenage years, whereas the frequency of ALL gradually decreased. The relative frequency of CML and MDS apparently started to increase in patients in their late teens. The relative frequency of M3 and t(15;17) gradually increased during adolescence. Among patients aged 1 to 4 years, M7 was the most frequent FAB subtype. Among patients aged 5 to 9 years, M2 and t(8;21) was the most frequent subtype. The percentage of T cell ALL increased in patients 5 to 9 years old, reaching 31.2% in the 20- to 24-year-old age group. The percentage of patients with hyperdiploidy over 50 chromosomes was highest (17.0%) in patients aged 1 to 4 and decreased to 3.9% in the older teens. The percentage of patients with the Ph1 chromosome increased from 9.9% in teens to 30.0% in patients in their late twenties. When comparing event-free survival (EFS) rates for ALL according to age, the estimated 7-year EFS rate was highest for patients aged 1 to 9 years (65.9%) and intermediate for patients aged 10 to 15 years (48.4%). However, the EFS rate was significantly worse for patients aged 15 to 19 years (19.4%) and 20 to 29 years (17.0%) (P = 0.024). On the other hand, the EFS rate for AML decreased with increasing age, although without statistical significance. The overall survival rates are approximate among all age groups. The results of the study indicate that there are considerable variations in biologic features of leukemia between children and young adults. The prognosis for adolescent leukemia may be improved by introducing pediatric trials, which take into account the prognostic biological features.