The percentage of long-term survivors in acute myeloid leukaemia (AML) in the elderly does not exceed 10-15% of patients enrolled into clinical trials because of lower complete remission (CR) rates and higher incidence of relapse. However, few data are available as the treatment of elderly patients with relapsed disease is concerned. The aim of this study was of collecting data on criteria adopted for the treatment of these patients. A questionnaire was e-mailed to 32 haematologic institutions involved in the Gruppo Italiano per le Malattie Ematologiche dell'Adulto (GIMEMA) group. Questions to be addressed regarded: (1) per cent of relapsed elderly patients treated with aggressive salvage chemotherapy; (2) the selection criteria adopted for inclusion into intensive reinduction; (3) the specific treatment adopted; (4) the treatment given to patients not eligible for intensive salvage. Per cent of patients enrolled into aggressive salvage regimens varied from 10 to 80% (median 50%). The most frequent factor influencing the therapeutic choice was performance status (97%). Additional factors were age >70 years (44%) and duration of first CR (53%). Fludarabine including regimens were most frequently used as aggressive salvage therapy (59%), while gemtuzumab ozogamicin was adopted in various combinations at 11 out of 32 institutions (34%). For patients not eligible to aggressive therapy, the most frequent approach included hydroxyurea (59%). Low dose ARA-C (LDARA-C) was adopted at five centres: as single agent (n = 1), with 6-thioguanine (n = 1), with vitamin D3 and all-trans retinoic acid (ATRA) (n = 2), or with ATRA alone (n = 1). The FLT3 inhibitor CEP-701 was used at one centre. We conclude that the treatment of AML in elderly relapsed patients is extremely heterogeneous. A marked selection is operated as to inclusion into aggressive salvage regimens and only a small minority of patients are offered experimental approaches.