The development of biomarkers able to predict the occurrence of nosocomial infection could help manage preventive strategies, especially in medical patients whose degree of acquired immunosuppression may be variable. We hypothesized that the NO fraction present in the airways (upper and lower) of critically ill patients under mechanical ventilation could constitute such a biomarker. We conducted an observational proof-of-concept study in a medical intensive care unit of a teaching hospital. Forty-five patients (26 men; 72 [25th-75th percentiles] years [56-82]; Simplified Acute Physiology Score II, 63 [50-81], 14 infected) under mechanical ventilation (>3 days) underwent on day 1 and day 3 of their stay: nasal and exhaled (partitioned in bronchial and alveolar sources) bedside NO measurements, determination of urine NO end products and plasma cytokine (IL-6, IL-10) concentrations, and Sequential Organ Failure Assessment score calculation. Nosocomial infection incidence was recorded during the 15 subsequent days. Fifteen patients (33%) acquired a nosocomial infection (16 infections, 15 ventilator-associated pneumonia and 1 bacteremia). Nasal NO was the only marker significantly different between patients with and without subsequent infection (day 1, 52 ppb [20-142] vs. 134 [84-203], P = 0.038; day 3, 98 ppb [22-140] vs. 225 [89-288], P = 0.006, respectively). Nasal NO fraction 148 ppb or less at day 3 had an 80% sensitivity, a 70% specificity, and an odds ratio of 2.7 (95% confidence interval, 1.9-3.8) to predict acquisition of nosocomial infection. Nonsurvivors had a higher IL-6 concentration on day 3 (P = 0.014), whereas their nasal NO fractions were not significantly different. Nasal NO seems to be a relatively sensitive and specific biomarker of subsequent nosocomial infection acquisition (at least for ventilator-associated pneumonia), which warrants confirmation in a multicenter trial.