The ability of glucocorticoids (GCs) to kill lymphoid cells led to their inclusion in essentially all chemotherapy protocols for lymphoid malignancies, particularly childhood acute lymphoblastic leukemia (ALL). GCs mediate apoptosis via their cognate receptor and subsequent alterations in gene expression. Previous investigations, including expression profiling studies with subgenome microarrays in model systems, have led to a number of attractive, but conflicting, hypotheses that have never been tested in a clinical setting. Here, we present a comparative whole-genome expression profiling approach using lymphoblasts (purified at 3 time points) from 13 GC-sensitive children undergoing therapy for ALL. For comparisons, expression profiles were generated from an adult patient with ALL, peripheral blood lymphocytes from GC-exposed healthy donors, GC-sensitive and -resistant ALL cell lines, and mouse thymocytes treated with GCs in vivo and in vitro. This generated an essentially complete list of GC-regulated candidate genes in clinical settings and experimental systems, allowing immediate analysis of any gene for its potential significance to GC-induced apoptosis. Our analysis argued against most of the model-based hypotheses and instead identified a small number of novel candidate genes, including PFKFB2, a key regulator of glucose metabolism; ZBTB16, a putative transcription factor; and SNF1LK, a protein kinase implicated in cell-cycle regulation.