6-Mercaptopurine is extensively used in the treatment of childhood lymphoblastic leukaemia to prolong the duration of remission achieved with other drugs. The response to remission maintenance therapy varies widely. We investigated the relationship between red blood cell 6-thioguanine nucleotide, a metabolite of 6-mercaptopurine, and myelosuppression in 22 children with acute lymphoblastic leukaemia in remission. The peripheral neutrophil count was used as an index of myelosuppression. 6-Mercaptopurine dose was related to 6-thioguanine nucleotide concentration (r = 0.4; P less than 0.001; n = 90; y = 18.51 + 0.36 x). Large individual variations around the regression line are observed. Neither 6-mercaptopurine dose nor 6-thioguanine nucleotide concentration was related to the neutrophil count at the time of sampling (day 0) or 7 days later. Both 6-mercaptopurine dose and 6-thioguanine nucleotide concentration correlated with the neutrophil count at day 14 (r = -0.33; P less than 0.01; n = 90 and r = -0.3; P less than 0.01; n = 90 respectively). This delay is compatible with a cytotoxic action on bone marrow stem cells. Excluding children with other, uncontrolled, potentially myelosuppressive influences the correlation between 6-thioguanine nucleotide concentration and neutropenia improved (r = -0.6; P less than 0.001; n = 37). A significant degree of neutropenia was observed by day 14 if the 6-thioguanine nucleotide concentration (day 0) was greater than 210 pmol/8 X 10(8) RBCs. The assay of 6-thioguanine nucleotide may highlight those individuals with pharmacokinetic resistance. Two children on continuous high dose 6-mercaptopurine, had low red blood cell 6-thioguanine nucleotide concentrations and neutropenia was not observed.