Bone marrow samples collected from acute myeloblastic leukemia (AML) patients in complete clinical and hematological remission were studied for the persistence of cytogenetic abnormalities. AML patients from the three favorable cytogenetic categories [inv 16, t(8;21) and t(15;17)] and patients from the unfavorable cytogenetic categories (+8, -5, -7 and Philadelphia-positive) were studied. Seventy-one patients had evaluable metaphase spreads in remission marrows and 20 (28%) had one or more abnormal metaphases identical to that present in the pretreatment marrow. All 20 of these patients relapsed within 78 weeks, thus there were no false positive studies. Fifty-one patients had only diploid metaphases in their complete remission marrow, 25 relapsed, and 21 remained in continuous complete remission. Thus there was a 49% false negative rate of this study. These data indicate that the failure to detect residual chromosomally abnormal cells in the bone marrow does not guarantee continuous complete remission. Cytogenetic study was most useful in the favorable cytogenetic groups and least useful in the unfavorable groups. The persistence of normal metaphases in pretreatment marrows did not affect outcome or risk of recurrence. Twenty-five of 34 evaluable patients who relapsed after remission had either the identical cytogenetic abnormality present in the pretreatment marrow or showed the identical abnormality with additional chromosomal changes. Thus study indicates that cytogenetic examinations of complete remission bone marrow samples in patients with AML provides an objective method for detecting residual leukemia, and identifies patients with a potential for prolonged disease-free survival.