BACKGROUND: Mild cognitive impairment (MCI) is an important prodromal stage of Alzheimer's disease (AD), affecting 69 million individuals worldwide. At present, there is a lack of a community-applicable tool for MCI screening. Exhaled breath volatile organic compounds (VOCs) have been used to distinguish MCI from cognitively normal (CN) individuals only in small sample size studies and the efficacy has not been compared with blood biomarkers. OBJECTIVE: This diagnostic study aimed to assess the feasibility of using exhaled breath VOCs detection by a portable micro-gas chromatography (μGC) device as a screening tool to discriminate MCI from CN individuals in a community population. METHODS: A detection model was developed and optimized from five distinct machine learning algorithms based on the differential VOCs between 240 MCI and 241 CN individuals. Among these 481 participants, five plasma biomarkers were measured in 397 individuals (166 MCI and 231 CN). RESULTS: The final model (481 individuals) incorporating eight differential VOCs showed good performance with an area under the receiver-operating characteristic curve (AUC) of 0.84 (95% confidence interval (95% CI): 0.83-0.85). The AUC of the VOC model (0.80, 95% CI: 0.69-0.90) was higher than that of the plasma model (0.77, 95% CI: 0.65-0.88) (397 individuals). CONCLUSIONS: The detection of exhaled breath VOCs by a portable μGC device is feasible for MCI screening in community populations, potentially facilitating early detection and intervention strategies for individuals at high risk.