To investigate the effect of variation components in flood series on flood frequency analysis of the Pinglu Canal, taking the hydrological control section of the Qinjiang River of the Pinglu Canal Project as an example, the empirical mode decomposition model was applied to extract variation components from the flood series. The effects of these components on flood frequency analysis were assessed through flood serial reconstruction and time-frequency domain analysis. Wavelet coherence analysis and physical cause exploration were further employed to investigate underlying causes. The results indicate that the annual maximum 1-d flood discharge of the Qinjiang River shows a decreasing trend, with a period of 3 to 21 a. The periodic variation components exhibit similarities with the patterns of El Nio, La Nia, and sunspot activities, while the decreasing trend may be attributed to environmental changes. The variation components alter the mean, kurtosis, and skewness characteristics of the flood series, serving as the statistical cause for differences in flood peak estimates under different design frequencies, and the inherent characteristics of El Nio and La Nia were identified as potential causes. The trend variation component has a greater influence on flood frequency analysis compared to other components, causing deviations of 0.16% to 18.81% in design flood peaks under the design frequency ranging from 0.1‰ to 50‰. Therefore, attention should be placed on the trend variation component and its influencing factor of environmental changes in flood frequency analysis.