Eosinophil count in nasal fluid (ECNF) was used to differentiate nasal pathologies. Receiver operating characteristic (ROC) curve analysis and the area under the curve (AUC) were performed to evaluate the ECNF's accuracy in distinguishing allergic rhinitis (AR) from non-allergic rhinitis (NAR). We also evaluated the accuracy of ECNF in recognizing patients with mild and severe symptoms of rhinitis and patients with ineffective and effective clinical responses to antihistamines. 1,170 consecutive adult patients with a clinical history of rhinitis were studied. ECNF's median in AR was 6.0 and 2.0 in NAR and the best cut-off value was > 3.0, AUC = 0.75. ECNF's median in AR with mild nasal symptoms was 3.0 and 7.0 with severe symptoms, and the best cut-off value was 4.0, AUC = 0.90. ECNF's median in NAR with mild nasal symptoms was 2.0 and 8.5 with severe symptoms, and the best cut-off value was > 4.0, AUC = 0.86. ECNF's median in AR with effective clinical response to antihistamines was 4.0 and 8.0 with ineffective response, the best cut-off value was < or = 5.0, AUC = 0.94. ECNF's median in NAR with an effective clinical response to antihistamines was 1.0 and 2.0 with ineffective response, and the best cut-off value was < or = 3.0, AUC = 0.64. Our results suggest an interesting practical use of ECNF data as evaluator of the clinical severity both AR and NAR. As predictor of the clinical response to antihistamines, ECNF is accurate only in patients with AR. The ECNF's performance was moderately accurate in distinguish patients with AR and NAR.