5-Azacytidine (AZA) is an hypomethylating agent approved in US for the treatment of all FAB subtypes of myelodysplastic syndromes (MDS). Some recent reports have raised the question of a possible efficacy of AZA in selected patients with acute myeloid leukaemia (AML). In September 2007, we started a retrospective study aiming to register and analyse all Italian patients with MDS or AML who had received AZA for the treatment of their disease outside of clinical trials, on the basis of a national patient named program. Among a total of 246 patients treated in 31 different Italian Institutions since 2005, 55 AML diagnosed according to WHO criteria were collected. Median age was 72 years (range 29–87) and 28 patients were male. Poor karyotype was present in 11 patients (20%), while 14 patients (25%) had secondary AML. Median time from diagnosis was 5 months (range 0–72). Eighteen patients (33%) received AZA as front-line treatment, as they were considered not eligible for intensive chemotherapy due to age, co-morbidities or poor performance status. Thirty-seven patients (67%) were pre-treated with growth factors (3 patients) or with one or more lines of chemotherapy (11 and 23 patients, respectively); most of the pre-treated patients (22 out of 34) had received high dose chemotherapy, including autologous or allogeneic stem cell transplantation. Low dose chemotherapy had been employed in the remaining cases. The median number of monthly AZA cycles administered was 4 (range 1–22). Thirty-nine patients (71%) received AZA at the fixed dose of 100 mg/d s.c., 16 patients (29%) received a dose of 75 mg/sqm/d s.c.. A seven-day per month schedule was employed in 43 patients (78%), while 11 patients (20%) received AZA for more than 7 days and one patient for 5 days. Twenty-nine patients (52.8%) received AZA alone, twenty-six (47.2%) in various combinations with growth factors (1), valproic acid +/− ATRA (21) or gentuzumab-ozogamycin (4).The most relevant toxicities observed (grade 3–4) were represented by further myelosuppression (15%), infections (24%: in particular, 1 fungal lung infection, 3 pneumonia and 1 septic shock) and gastrointestinal adverse events (20%). The overall response rate was 35.3% (18/51): 8 patients achieved a complete remission (CR) (15.7%), while a partial response (PR) was observed in 5 patients (9.8%). Five haematological improvements were also seen (9.8%). Response rate was significantly higher in untreated patients compared to pre-treated ones (p=0.02). A statistically significant difference (p=0.04) in response rate in favour of 75 mg/sqm/d versus 100 mg fixed dose was also observed. The actuarial probability of overall survival (OS) at 16 months was 45% for patients responding to AZA and 10% for those non responding (p=0.0027). In conclusion, our data show that: AZA can induce significant responses in about one third of AML patients; the “standard” dose of 75 mg/sqm/d seems to be more effective than 100 mg/d (one single vial) fixed dose; AZA is more effective in de novo as compared to pre-treated (refractory and/or relapsed) disease; AML patients responding to AZA have a significant survival advantage compared to non responders.