This study is based on a prospective cohort of 33,987 young and middle-aged adults aged 18 to 60, exploring the predictive value of cumulative plasma atherogenic index (cumAIP) for the risk of non-alcoholic fatty liver disease (NAFLD). By calculating the weighted average AIP value from multiple follow-ups (i.e., cumAIP), it more truly reflects the long-term exposure level of lipid metabolism abnormalities. The study found that the higher the cumAIP, the greater the risk of NAFLD, showing a clear dose-response relationship, with the Q4 group's risk increasing by 79% compared to the Q1 group. The innovation lies in the first-time proposal and verification of cumAIP being superior to single AIP, which can more accurately predict the risk of NAFLD, and the introduction of time-accumulated exposure models and competing risk analysis, enhancing the robustness of the results.