The development of new treatments for older patients with acute myeloid leukaemia (AML) is an active area, but has met with limited success. Sapacitabine is a novel orally administered nucleoside analogue that has shown encouraging activity in unrandomised early-stage trials. We randomised 143 untreated patients with AML or with high-risk myelodysplastic syndrome (>10% marrow blasts) between sapacitibine and low-dose ara-C (LDAC) in our 'Pick a Winner' trial design. At the planned interim analysis there was no difference between LDAC and sapacitibine in terms of remission rate (CR/CRi, 27% vs 16% hazard ratio (HR) 1.98(0.90-4.39) P=0.09), relapse-free survival (10% vs 14% at 2 years, HR 0.73(0.33-1.61) P=0.4) or overall survival (OS; 12% vs 11% at 2 years, HR 1.24(0.86-1.78) P=0.2). Sapacitibine was well tolerated, apart from more grade 3/4 diarrhoea. On the basis of these findings sapacitibine did not show sufficient evidence of benefit over LDAC for the trial to be continued.