Although central nervous system (CNS) leukemic relapse is frequent in adult acute lymphocytic leukemia (ALL), the need for prophylaxis in different risk groups for CNS relapse, the value of high-dose systemic and intrathecal (IT) chemotherapy, and the timing of prophylaxis are not well defined. This analysis was conducted to investigate these questions and to assess the value of a risk-oriented CNS prophylaxis approach. We analyzed the incidence of CNS leukemia after initiation of therapy in patients treated on 4 consecutive trials for adult ALL including different CNS prophylactic modalities. The treatment groups included (1) the program preceeding the vincristine-Adriamycin- dexamethasone (VAD) regimen, with no CNS prophylaxis; (2) the VAD regimen with prophylaxis using high-dose systemic chemotherapy; (3) the modified VAD program with high-dose systemic chemotherapy to all patients and IT chemotherapy for high-risk patients after achieving complete remission; and (4) the hyperCVAD program with early high-dose systemic and IT chemotherapy starting during induction to all patients, with more IT injections (16IT) administered to the high-risk group for CNS relapse compared with the low-risk group (4IT). A total of 391 patients were included, 73 of whom were treated with preVAD, 112 with VAD, 114 with modified VAD, and 92 with hyperCVAD. The overall CNS relapse rates were 31%, 18%, 17%, and 3%, respectively for the 4 groups (P < .001). For the high-risk group for CNS relapse, they were 42%, 26%, 20%, and 2%, respectively (P < .001). The differences in CNS relapse rates in the low-risk group were not statistically significant. At 3 years, the overall CNS leukemia event-free rates were 48%, 76%, and 98%, respectively (P < .001). In the high-risk group, the CNS event- free rates were 38%, 66%, 75%, and 98%, respectively (P < .001); however, there was no difference in the low-risk group. We conclude that (1) high-dose systemic chemotherapy is a useful prophylactic measure; (2) early IT chemotherapy is necessary to reduce the incidence of CNS leukemia overall and in the high-risk group; and (3) a risk- oriented approach is appropriate to tailor the intensity of CNS prophylaxis.