This open-label, prospective, observational study aimed to evaluate disease-free survival (DFS), overall survival (OS), PML-RARα polymerase chain reaction (PCR) monitoring and safety in elderly patients with de novo acute promyelocytic leukemia (APL) who were treated with either arsenic trioxide (As2O3) or medium-dose cytosine arabinoside (MiDAC) as frontline consolidation regimens. A total of 167 patients (age≥65 years old) received all-trans retinoic acid + daunorubicin as induction therapy. Of these patients, 22 died before attaining complete remission; the remaining 145 subjects received MiDAC- or As2O3-based consolidation therapy. As2O3 was superior to MiDAC for improving DFS and OS. This benefit appeared to result from the longer 5-year DFS (92.0 vs. 69.1%, P<0.01) and OS (94.5 vs. 79.7%, P<0.05) of As2O3 compared to MiDAC. PCR monitoring demonstrated that As2O3 promoted a lower positive rate than MiDAC (21.7 vs. 4.5%, P<0.05), but this treatment had no advantage for maintaining a low positive rate in the high-risk group. The most common life-threatening adverse drug effects in patients with MiDAC were platelet counts<25×10(9)/L (85.7%), leukocyte counts<1.0×10(9)/L (81.4%) and severe infection (84.3%). In contrast, the As2O3 regimen rarely caused leukocyte counts<1.0×10(9)/L (22.7%, P<0.01), platelet counts<25×10(9)/L (37.3%, P<0.01) or severe infection (21.3%, P<0.01). These data confirm that MiDAC should not be added during the initial consolidation of patients with APL because this treatment is far less effective, particularly in patients with a low-risk profile, and far more toxic than As2O3.