Deletion of the 9p21 chromosomal region is frequently found in childhood acute lymphoblastic leukemia (ALL). The target of these deletions is CDKN2A, a gene encoding both p16(INK4a) and p14(ARF). However, contiguous genes such as CDKN2B, encoding p15(INK4b), or MTAP, encoding methylthioadenosine phosphorylase, can be included in the deletions. Gene dosage by use of real-time PCR has recently been proposed as a promising technical option for the diagnosis of deletions. However, its reliability and its capacity to detect mono-allelic deletions in tumor samples are controversial. To evaluate the frequency and extent of deletions in 284 children with ALL, we devised a real-time PCR assay for CDKN2A, CDKN2B exons 1beta and 3, and MTAP gene dosage and validated it by comparison with loss-of-heterozygosity analysis. We show that, if several controls and adjustments are performed, real-time PCR can provide a reliable test for mono- and bi-allelic deletions in ALL. We propose a strategy that overcomes the major caveats of such a dosage in tumor samples: aneuploidy and contamination by normal cells. By use of this assay, we found bi-allelic deletions in 58 and 17% of T- and B-lineage ALL, respectively. Mono-allelic deletion was observed in about 15% of cases, stressing the importance of their detection in ALL. CDKN2B and/or MTAP co-deletions were highly variable in both T- and B-lineage ALL, making ALL with 9p21 a rather heterogeneous group. Because proteins encoded by these genes might influence the response to treatment, the prognosis of 9p21-deleted ALL could vary according to the extent of the deletion.