Résumé : Background: Plasma concentrations of endocan, a proteoglycan preferentially expressed in the pulmonary vasculature, may represent a biomarker of lung (dys)function. We sought to determine whether the measurement of plasma endocan levels early in the course of acute respiratory distress syndrome (ARDS) could help predict risk of death or of prolonged ventilation. Methods: All patients present in the department of intensive care during a 150-day period were screened for ARDS (using the Berlin definition). Endocan concentrations were measured at the moment of ARDS diagnosis (T0) and the following morning (T1). We compared data from survivors and non-survivors and data from survivors with less than 10 days of ventilator support (good evolution) and those who died or needed more than 10 days of mechanical ventilation (poor evolution). Results are presented as numbers (percentages), mean ± standard deviation or medians (percentile 25–75). Results: Ninety-six consecutive patients were included [median APACHE II score of 21 (17–27) and SOFA score of 9 (6–12), PaO2/FiO2 ratio 155 (113–206)]; 64 (67%) had sepsis and 51 (53%) were receiving norepinephrine. Non-survivors were older (66 ± 15 vs. 59 ± 18 years, p = 0.045) and had higher APACHE II scores [27 (22–30) vs. 20 (15–24), p < 0.001] and blood lactate concentrations at study inclusion [2.1 (1.3–4.0) vs. 1.5 (0.9–2.6) mmol/L, p = 0.024] than survivors, but PaO2/FiO2 ratios [150 (116–207) vs. 158 (110–206), p = 0.95] were similar in the two groups. Endocan concentrations on the day after ARDS diagnosis were significantly higher in patients with poor evolution than in those with good evolution [12.0 (6.8–18.6) vs. 7.2 (5.4–12.5), p < 0.01]. Conclusion: Blood endocan concentrations early in the evolution of ARDS may be a useful marker of disease severity.