To assess the frequency and significance of nonrandom abnormalities of chromosome 9p in childhood acute lymphoblastic leukemia (ALL), we analyzed our experience with 398 consecutive cases with completely banded karyotypes. Forty cases (10%) with abnormalities of 9p were identified: 26 with deletions, nine with unbalanced translocations resulting in the loss of 9p material, and five with apparently balanced reciprocal translocations. As compared with children with ALL lacking 9p abnormalities, these 40 cases were significantly older, had higher initial circulating WBC counts, more "lymphomatous" disease characteristics (including presence of a mediastinal mass in 15%. T-cell phenotype in 26%, splenomegaly greater than 8 cm in 25%), an increased failure rate in the first 2 to 3 years after diagnosis, and a higher incidence of extramedullary relapse. Conversely, lymphomatous ALL cases were twice as likely (19% v 8%) to have an abnormality of chromosome 9p than ALL cases lacking lymphomatous features (P = .01). The finding of an abnormal chromosome 9p, however, was not specific for lymphomatous ALL or T-cell lineage, because most cases were neither lymphomatous nor T-cell, and the overall Kaplan-Meier distribution of treatment failures for abnormal 9p cases was not statistically significantly different from control ALL cases receiving the same treatment who lacked abnormalities of 9p (P = .06, by log-rank test). We conclude that nonrandom abnormalities of chromosome 9p, especially a breakpoint in 9p21-22, occur with increased frequency in childhood ALL in association with some high-risk clinical features. Despite this association, the chromosome anomaly is nonspecific in its syndrome delineation and confers no major adverse consequence on long-term survival of childhood ALL treated with modern therapy. However, due to an apparently increased hazard of involvement of the CNS (eight of 17 failures), it may be inadvisable to lessen the intensity of CNS preventive therapy for this group of patients.