Abstract: To analyze the impacts of electric vehicle (EV) charging loads on distribution network harmonics, a frequency-domain analysis framework based on harmonic extended linear modeling (HELM) is established. The cross-frequency coupling between voltage and current harmonics is characterized using a frequency coupling matrix. Sampling at 20 kHz and adopting a steady-state window of three fundamental cycles, the spectrum is extracted via discrete Fourier transform. Indicators including total harmonic distortion (THD), total demand distortion (TDD), root-mean-square value, peak value and crest factor are calculated. Furthermore, kernel density estimation, empirical distribution, the Kolmogorov–Smirnov test, principal component analysis, k-means clustering and three-dimensional visualization are jointly employed to systematically compare two scenarios: "EV loads superimposed with other loads" and "other loads only". Simulation results reveal that the voltage distributions of the two scenarios are highly consistent, while discrepancies mainly emerge in current characteristics. For the scenario incorporating EVs, the current root-mean-square value, THD and crest factor are generally lower with smaller dispersion; the energy proportion of low-order harmonics (2nd–10th orders) decreases, demonstrating a more fundamental-wave-dominated spectral feature. The proposed research provides methodological support and quantitative references for distribution network harmonic mitigation, optimization of filtering and rectification schemes for charging facilities, and improvement of assessment criteria for power quality.