Elevated concentrations of soluble urokinase plasminogen activator receptor (suPAR) predict progression to severe respiratory failure (SRF) or death among patients with COVID-19 pneumonia and guide early anakinra treatment. As suPAR testing may not be routinely available in every health-care setting, alternative biomarkers are needed. We investigated the performance of C-reactive protein (CRP), interferon gamma-induced protein-10 (IP-10) and TNF-related apoptosis-inducing ligand (TRAIL) for predicting SRF or death in COVID-19. Two cohorts were studied; one discovery cohort with 534 patients from the SAVE-MORE clinical trial; and one validation cohort with 364 patients from the SAVE trial including also 145 comparators. CRP, IP-10 and TRAIL were measured by the MeMed Key® platform in order to select the biomarker with the best prognostic performance for the early prediction of progression into SRF or death. IP-10 had the best prognostic performance: baseline concentrations 2000pg/ml or higher predicted equally well to suPAR (sensitivity 85.0%; negative predictive value 96.6%). Odds ratio for poor outcome among anakinra-treated participants of the SAVE-MORE trial was 0.35 compared to placebo when IP-10 was 2,000pg/ml or more. IP-10 could divide different strata of severity for SRF/death by day 14 in the validation cohort. Anakinra treatment decreased this risk irrespective the IP-10 concentrations. IP-10 concentrations of 2,000pg/ml or higher are a valid alternative to suPAR for the early prediction of progression into SRF or death the first 14days from hospital admission for COVID-19 and they may guide anakinra treatment. gov, NCT04680949 and NCT04357366.