The response to chemotherapy is determined essentially by two factors: first, pharmacokinetic factors, determining which concentration of drug reaches the malignant cells, and second, cellular drug resistance of these cells, determining how many of them will be killed by that concentration of drug. The study of cellular drug resistance has been stimulated by the development of short-term 'total cell kill' assays, such as the MTT assay, for use on patient samples. The drug resistance profiles differed markedly between ALL and ANLL, between immunophenotypic and karyotypic subgroups within ALL, and between initial and relapsed ALL. The results of the MTT assay showed a significant relation between the antileukemic activity of prednisolone in vitro and the clinical response to systemic monotherapy with that drug. At multivariate analysis including several well-known prognostic factors (WBC, age, immunophenotype) only the in vitro resistance to prednisolone, dexamethasone, L-asparaginase, and daunorubicin was significantly related to clinical outcome. At multiple regression analysis, combination of the results for prednisolone, L-asparaginase, and vincristine made it possible to distinguish between three patient groups with increasing levels of drug resistance and markedly different probabilities of 2-year disease-free survival: 100%, 83%, and 60%. These results show that in vitro drug resistance testing can give a correct prediction of prognosis, superior to that of currently used prognostic factors. Stratification of prognostic groups based on the results of drug resistance testing is feasible and should be introduced into new clinical trials. Many questions now remaining could be answered within carefully designed preclinical and clinical studies.