Background: We aimed to determine whether treatment with oseltamivir was associated with decreased mortality among patients admitted to intensive care units (ICUs) with severe influenza and whether early therapy (within 48 hours of symptoms onset) was associated with improved survival rates.Methods: This was a prospective observational study of patients with confirmed influenza who were admitted to 184 ICUs in Spain. The primary outcomes were to investigate the association of ICU mortality with oseltamivir therapy by comparing patients who are untreated (UT group) with those who are treated (all treated; AT group) and those in early treatment (ET) and late treatment (LT) groups. We also assessed the associations of both ET and LT with ICU mortality and compared the results.Findings: We enrolled 4,175 patients of whom 3,537 met the inclusion criteria. Most patients were diagnosed with influenza pneumonia (84·7%). When comparing the AT (n=3,439) and UT (n=98) groups, the former had higher survival after adjusting for severity and confounding factors (Hazard ratio [HR] 0·67; p=0·03). We included 3,388 patients to compare the ET and LT groups. In the multivariable analysis, ICU mortality was lower in the ET group, having an odds ratio of 0·7 (p=0·004). When propensity score matching was applied to a matched cohort (ET, n=790; LT, n=2,522), the ET group had lower ICU mortality (HR 0·78; p=0·008) compared with the LT group, even after a competing risks analysis (sub-HR 0·79; p=0·01).Funding Statement: SEMICYUC (Spanish Society of Critical Care) and Ricardo Barri Casanovas Foundation.Declaration of Interests: JSN-V-T has received research funding from F. Hoffman-La Roche for separate work. He is currently seconded to the Department of Health and Social Care, England (DHSC). The views expressed in this paper are those of the authors and not necessarily those of DHSC. CG-V has received the INTENSIFICACIÓ Grant- a grant supported by the Catalan Health Agency [PERIS (Pla estratègic de recerca i innovació en salut – ‘Strategic Plan for Research and Innovation in HealthCare’)]. All other authors declare no competing interests.Ethics Approval Statement: The study was approved by the Joan XXIII University Hospital Ethics Committee (IRB#11809). All data were anonymised, allowing the requirement for informed consent to be waived.