Although the effects of meteorological factors on severe air pollution have been extensively investigated, quantitative decomposition of the contributions of meteorology and anthropogenic factors remains a big challenge. The novel coronavirus disease 2019 (COVID-19) pandemic affords a unique opportunity to test decomposition method. Based on a wind decomposition method, this study outlined an improved method to differentiate complex meteorological and anthropogenic effects. The improved method was then applied to investigate the cause of unanticipated haze pollution in China during the COVID-19 lockdown period. Results from the wind decomposition method show that weakened winds increased PM_2.5 concentrations in the Beijing–Tianjin area and northeastern China (e.g., by 3.19 μg/m³ in Beijing). Using the improved decomposition method, we found that the combined meteorological effect (e.g., drastically elevated humidity levels and weakened airflow) substantially increased PM_2.5 concentrations in northern China: the most substantial increases were in the Beijing–Tianjin–Hebei region (e.g., by 26.79 μg/m³ in Beijing). On excluding the meteorological effects, PM_2.5 concentrations substantially decreased across China (e.g., by 21.84 μg/m³ in Beijing), evidencing that the strict restrictions on human activities indeed decreased PM_2.5 concentrations. The unfavorable meteorological conditions, however, overwhelmed the beneficial effects of emission reduction, causing the severe haze pollution. These results indicate that the integrated meteorological effects should be considered to differentiate the meteorological and anthropogenic effects on severe air pollution.