BACKGROUND: The role of exhaled H 2 S as a marker of airway inflammation and its relationship with COPD severity remain to be determined. METHODS: Airway inflammation was classified in 77 COPD subjects based on the presence of inflammatory cells in induced sputum. We investigated the association between disease phenotype and exhaled H 2 S, lung function, and plasma levels of several inflammatory factors, including tumor necrosis factor alpha, interleukin-8, and leukotriene B 4 . RESULTS: In total, 33.77% of enrolled COPD subjects were diagnosed with eosinophilia. These subjects had a longer disease course, smoked fewer cigarettes, and experienced more frequent exacerbation events before study enrollment. However, they also had worse lung function and larger residual volume, they demonstrated greater changes in FEV 1 following bronchodilator inhalation. Although levels of plasma inflammatory factors did not significantly differ between subjects with and without eosinophilia, subjects without eosinophilia had significantly higher levels of exhaled H 2 S (9.19 ؎ 2.74 vs 7.24 ؎ 1.68 parts per billion, P ؍ .01). Furthermore, exhaled H 2 S levels were negatively correlated with induced sputum eosinophils (r ؍ ؊0.45, P ؍ .05), and positively correlated with inspiratory capacity in COPD subjects (r ؍ 0.51, P ؍ .026), but did not correlate significantly with plasma inflammatory factors. A cut-off value of 7.10 parts per billion of exhaled H 2 S predicted a non-eosinophilic phenotype with 68.6% sensitivity and 77.9% specificity. CONCLUSIONS: Exhaled levels of H 2 S were lower in subjects with eosinophilia. Increased levels of exhaled H 2 S predicted a non-eosinophilic phenotype in our study population. Key words: chronic obstructive pulmonary disease; airway inflammation phenotype; eosinophilia; hydrogen sulfide. [Respir Care 2015;60(2):251-258.