VAP occurred in nine patients and the 19 patients who did not develop VAP were considered controls. There were no differences between patients with VAP and controls with respect to age, gender, initial APACHE II score, and primary diagnosis. Levels of TNF, IL-10, IL-6 and IL-8 did not change in control patients in either plasma or NBL-fluid. Furthermore, the diagnosis of VAP was not associated with changes in plasma cytokines. However, serial changes in TNF, IL-10, IL-6 and IL-8 in NBL-fluid strongly correlated with the diagnosis of VAP. A rise of TNF in NBL-fluid above 200 pg/ml predicted a 4.0 (95% CI: 1.1–15.1) times increased risk for developing VAP (P = 0.04, time-dependent Cox proportional hazard analysis). An increase of IL-10, IL-6 and IL-8 levels in NBL-fluid above 100 pg/ml, 1 ng/ml, and 15 ng/ml, respectively, was associated with a relative risk of 5.6 (95% CI: 1.5–20.9), 9.0 (95% CI: 1.1–72.1), and 4.6 (95% CI: 0.9–22.6), respectively, for developing VAP.