Single-valued neutrosophic set (SVNS) is very suitable for expressing indeterminate and inconsistent information in fault diagnosis problems, and then its cosine measure is a useful mathematical tool for handling the decision making, pattern recognition, and fault diagnosis problems. However, due to the lack of engineering applications of SVNSs in fault diagnoses, the paper develops a cosine measure-based fault diagnosis method and applies it to the misfire fault diagnosis of gasoline engines with SVNS information. Through the cosine measure between each fault pattern and a real-testing sample, according to the largest cosine measure value, we can determine that the testing sample should belong to the fault pattern. Finally, we provide nine real-testing samples to illustrate the misfire fault diagnoses of gasoline engines. All diagnosis results are in accordance with actual fault types. The results demonstrate the effectiveness and rationality of the proposed diagnosis method