Reliable clinical or molecular predictors of benefit from azacitidine therapy in patients with myelodysplastic syndromes (MDS) are not defined. Doubling of platelet count at start of second cycle of azacitidine therapy compared to baseline was associated with achieving response and survival advantage in a Dutch cohort. To validate this observation, we analysed a larger cohort of North American patients, whose data was collected in a prospective clinical trial with a longer median follow-up. We found a significant association between platelet count doubling after first cycle of azacitidine therapy and probability of achieving objective response. Among patients with MDS or oligoblastic acute myeloid leukaemia (<30% bone marrow blasts, n = 102), there was a statistically significant reduction in risk of death for patients who achieved platelet count doubling (n = 23, median OS, 21·0 months) compared to those who did not (n = 79, median OS, 16·7 months, adjusted hazard ratio (no/yes)=1·88, 95% confidence interval, 1·03-3·40, P = 0·04). Nonetheless, the addition of this platelet count doubling variable did not improve the survival prediction provided by the revised International Prognostic Scoring System or the French Prognostic Scoring System. Identification of reliable and consistent predictors for clinical benefit for azacitidine therapy remains an unmet medical need and a top research priority.