Update informaiton. Update information. Asymptomatic infection has become one of the biggest challenges in controlling the spread of COVID-19. However, the rapid diagnosis and pathogenesis elucidation of the disease remain critical issues. Here, we performed comprehensive lipidomic profiling of 267 serum samples from 89 asymptomatic COVID-19 patients and 178 healthy controls. A panel of 15 key lipids was discovered that could high-efficiently distinguish asymptomatic patients by using a new ensemble learning model, which provided a high accuracy of 96.0% with only 3.6% false positive rate and 4.8% false negative rate. In addition, characteristic lipidomic changes revealed distinct dysregulated lipid metabolism upon asymptomatic SARS-CoV-2 infection, which implicated the specific host immune, inflammatory, and antiviral responses during the virus invasion in asymptomatic COVID-19. This study provides an effective method for the pre-diagnosis of asymptomatic COVID-19, and the revealed molecular signature provided clues for the pathogenesis and therapy of this disease.