Summary To define a prognostic model for predicting outcome of reduced‐intensity allogeneic stem cell transplantation (RIC‐ASCT) for myelofibrosis we evaluated 150 homogenously treated patients and developed a new risk score for overall survival (OS). In a multivariate Cox model for OS, only JAK2 V617F wild‐type, age ≥57 years and constitutional symptoms were independently predictive for OS (Hazard Ratio: 2·02; 2·43 and 2·80 respectively). Depending on the presence of one, two or all of these factors, HR of death was 3·08; 4·70 and 16·61 respectively (P < 0·001). This score was compared to the Lille, Cervantes, International Prognostic Scoring System (IPSS), dynamic IPSS (DIPSS) and modified European Blood and Marrow Transplantation Group (EBMT) scores. Lille score correlated significantly with OS but discriminated poorly between the intermediate and high‐risk groups (5‐year OS 56% and 51% respectively). IPSS and DIPSS correlated significantly with OS but differences between intermediate‐1 and intermediate‐2 groups were not significant (5‐year OS 78% vs. 78% and 70%, 60% respectively). Modified EBMT and Cervantes models did not predict OS post‐ASCT. In conclusion, a simple model which includes: age, JAK2 V617F‐status and constitutional symptoms can clearly separate distinct risk groups and can be used in addition to the Lille model to predict OS after RIC‐ASCT for myelofibrosis.