Anemia is a significant problem in patients on ICU. Its commonest cause, iron deficiency (ID), is difficult to diagnose in the context of inflammation. Hepcidin is a new marker of ID. We aimed to assess whether hepcidin levels would accurately guide treatment of ID in critically ill anemic patients after a prolonged ICU stay and affect the post-ICU outcomes. In a controlled, single-blinded, multicenter study, anemic (WHO definition) critically ill patients with an ICU stay ≥ 5 days were randomized when discharge was expected to either intervention by hepcidin treatment protocol or control. In the intervention arm, patients were treated with intravenous iron (1 g of ferric carboxymaltose) when hepcidin was < 20 μg/l and with intravenous iron and erythropoietin for 20 ≤ hepcidin < 41 μg/l. Control patients were treated according to standard care (hepcidin quantification remained blinded). Primary endpoint was the number of days spent in hospital 90 days after ICU discharge (post-ICU LOS). Secondary endpoints were day 15 anemia, day 30 fatigue, day 90 mortality and 1-year survival. Of 405 randomized patients, 399 were analyzed (201 in intervention and 198 in control arm). A total of 220 patients (55%) had ID at discharge (i.e., a hepcidin < 41 μg/l). Primary endpoint was not different (medians (IQR) post-ICU LOS 33(13;90) vs. 33(11;90) days for intervention and control, respectively, median difference - 1(- 3;1) days, p = 0.78). D90 mortality was significantly lower in intervention arm (16(8%) vs 33(16.6%) deaths, absolute risk difference - 8.7 (- 15.1 to - 2.3)%, p = 0.008, OR 95% IC, 0.46, 0.22-0.94, p = 0.035), and one-year survival was improved (p = 0.04). Treatment of ID diagnosed according to hepcidin levels did not reduce the post-ICU LOS, but was associated with a significant reduction in D90 mortality and with improved 1-year survival in critically ill patients about to be discharged after a prolonged stay. www.clinicaltrial.gov NCT02276690 (October 28, 2014; retrospectively registered).