The aim of this study was to investigate clinical parameters that might predict complete remission (CR) following reinduction therapy in patients with acute myeloid leukemia (AML). We retrospectively analyzed outcomes in 142 patients who failed to achieve CR with standard anthracycline-plus-cytarabine induction chemotherapy (IND1) and who received the same regimen as a reduction therapy (IND2). The CR rate after reinduction was 54%. Multivariate analysis showed that the presence of peripheral blood (PB) blasts on the day of commencement of IND2 and ≥20% blasts in an interim BM sample taken during IND1 (C1-INT-BM) were independent risk factors for failure of remission. Our scoring system for prediction of CR after reinduction (CRAR score) included a favorable chromosome risk group, absence of PB blasts on the day of commencement of IND2, and ≤21% blasts in the C1-INT-BM. This scoring system predicted that the rates of CR in the four subgroups would be 92.1%, 68.9%, 29.5%, and 7.3%, respectively, in good agreement with observed CR rates. The CRAR scoring model might be associated with the possibility of achieving CR after reinduction, and may be helpful in choosing alternative strategy for AML patients refractory to standard induction chemotherapy, although further prospective validation is required.