The expression of the DNA excision repair enzyme uracil-DNA glycosylase was investigated in bone marrow and peripheral samples from seven patients with acute lymphoblastic leukemia (ALL), from 17 patients with acute non-lymphocytic leukemia (ANLL), and from one patient with chronic granulocytic leukemia (CGL) in blast crisis. In addition, uracil-DNA glycosylase activities were determined in nine human leukemia/lymphoma cell lines. There was a clear correlation between the percentage of blast cells and the enzyme activity when mononuclear cell fractions from patient samples were analysed. The following uracil-DNA glycosylase activities were recorded (mean +/- S.D., number of samples): ALL = 45.6 +/- 14.8 U/mg of protein, N = 10; ANLL = 41.1 +/- 13.8 U/mg of protein, N = 22; CGL (blast crisis) = 44.7 U/mg of protein. The uracil-DNA glycosylase activity in nine human leukemia/lymphoma cell lines ranged from 35.2 to 66.0 U/mg of protein, and no striking differences were observed between the T-ALL, B-ALL, null cell ALL or myeloid lines. Similarly, the various biological features, such as the common ALL surface antigen, the terminal deoxynucleotidyl transferase enzyme, the sub-type of leukemia, chromosomal aberrations, or previous chemotherapy, did not apparently affect the expression of uracil-DNA glycosylase. We propose that the integrity of the genetic information is well protected by uracil-DNA glycosylase in different forms of leukemia, including cases with a low proportion of S-phase blasts, as assessed by flow cytometry in the present work. When compared to the activities in benign hematopoietic progenitor cells, studied previously in this laboratory, no big differences between the benign and malignant hematopoiesis were demonstrated. Hence, it is unlikely that selectivity of chemotherapy towards malignant vs benign hematopoietic growth could be based on the enzyme uracil-DNA glycosylase.