The efficacy of chemotherapy in acute myeloid leukaemia (AML) is limited by clinical drug resistance. We determined in vitro resistance to cytosine arabinoside (ARAC), daunorubicin (DNR), mitoxantrone (MITOX), m-amsacrine (AMSA) and etoposide (VP16) in 49 adults with de novo AML using the MTT assay. Results showed that nonresponders to chemotherapy were, in vitro, 2.9-fold more resistant to DNR, but not more resistant to ARA-C, compared to complete responders. However, complete responders who were in vitro resistant to ARA-C had a 4-fold higher risk of relapse (95% CI 1.3-12.5-fold) compared to complete responders in vitro sensitive to ARA-C. With a mean follow-up of 12 months the probability of continuous complete remission (CCR) for patients in vitro sensitive to ARA-C was 61% at 34 months (95% CI 28-82%), whereas all patients in vitro resistant to ARA-C relapsed within 18 months from diagnosis. This difference appeared to be independent of other clinical features such as sex, age, white blood cell count, FAB classification, and CD34 expression. In vitro resistance to DNR was not related to the probability of CCR. We conclude that in vitro drug resistance assessed with the MTT assay appears to be associated with short- and long-term clinical outcome in AML. Confirmatory studies comprising a sufficient number of patients for multivariate analyses should prove whether in vitro resistance to ARA-C will appear to be an independent risk factor.