The sensitivity of relapsed acute lymphoblastic leukemia (ALL) to treatment with steroids was assessed by measuring bone marrow blast, whole-cell glucocorticoid receptor (GR) levels and serial cytokinetic and clinical responses to steroids, both as single agents and in combination reinduction chemotherapy. GR levels ranged widely, from 1800 to 47,800 sites/cell (median 16,000), and did not differ significantly from levels measured in newly diagnosed patients (p = 0.50). Nineteen of the 48 children studied had GR levels measured both at diagnosis and relapse, and in 14 the values at relapse were either decreased or increased by more than 25%. For nine children, the sensitivity to a three-day oral course of dexamethasone alone (6 mg/m2 per day) was estimated from daily leukocyte counts and serial bone marrow cytokinetic studies. A clinical oncolytic effect of dexamethasone was associated with a reduction in the marrow cell proliferating compartment, as judged from labeling indices, mitotic indices and percent S + G2 + M in five patients. We found, however, that GR levels either alone or combined with clinical or cytokinetic responses to dexamethasone as a single agent did not reliably identify patients who were likely to respond favorably to reinduction with steroid-containing combination chemotherapy. The 35 children who successfully attained subsequent remissions had GR levels similar to those who failed (p = 0.35). While not statistically significant, two observations suggest an association between lower receptor content and drug resistance. Receptor levels from patients with ALL who relapsed while on chemotherapy were appreciably lower (median 15,700, n = 31) than GR levels from patients who relapsed off therapy (median 20,300, n = 17) (p = 0.08). Moreover, three of seven patients with serial studies, whose GR levels at relapse were lower than at diagnosis, failed reinduction--compared to only one of twelve whose levels were either increased or remained unchanged (p = 0.11). Although not having any obvious clinical utility, studies of GR at diagnosis and at relapse may aid in clarifying mechanisms of drug resistance in leukemic blasts.