Abstract: The variations of daily PM_2.5 concentration in the two capital cities (Hefei and Nanjing) in western Yangtze River Delta district are highly correlated. To investigate the impact of transport pattern on PM_2.5 concentration in this area, the cluster analysis was used to categorize the daily 72 h back trajectories of Hefei at 100 m above ground level (AGL), which represents near surface, and 1000 m AGL, which represents the mid-high level of the boundary layer, on days with rainfall lower than 10 mm during 3 a period from 2013 to 2015. The back trajectories were divided into seven groups at 100 m and six groups at 1000 m. The relationship between PM_2.5 concentration and the transport pattern was studied based on the results of cluster analysis in combination with daily averages of PM_2.5 concentration, horizontal visibility, surface wind speed and relative humidity at ground level. The results are as follows: (1) The statistical results of PM_2.5 concentration, visibility, wind speed, and relative humidity in different clusters were evidently different at both 100 and 1000 m levels. (2) At 100 m, the highest cluster-mean PM_2.5 concentration, which was almost double of the lowest value, the severe PM_2.5 pollution (daily mean PM_2.5 concentration >150 μg/m³) and the lowest daily average visibility were found in the cluster with the shortest cluster-mean trajectory mainly coming from the east. More than 60% of total severe pollution days fell within this cluster, which accounted for about 30% of total days (the biggest percentage among all clusters). The air mass in this cluster moved to the studied area with very weak descending motions during the past 72 h, especially in those severe pollution days, the averaged descending height was only 28 m. The second highest daily mean PM_2.5 concentration and the number of severe PM_2.5 pollution days fell within the cluster with short trajectories from the northwest. The trajectories in this cluster accounted for 14% of the total. In thisv cluster, the air mass moved to the studied area with evident downward motion, indicating that the long-range transport of pollutants intensified local PM_2.5 pollution. According to daily changes in PM_2.5 concentration, the above two clusters usually corresponded to increasing daily average PM_2.5 concentration. The two clusters with the lowest cluster-mean PM_2.5 concentration had long trajectories from the northeast and southwest, which accounted for 6.4% and 10.3% of the total. They corresponded to decreasing daily average PM_2.5 concentrations. (3) The results of statistics with clustering of trajectories at 1000 m were similar to those at 100 m. However, the differences in PM_2.5 concentration among clusters were smaller than those at 100 m, and different from those for PM₁₀ at the beginning of the 2000s. (4) The back trajectories on 84 severe PM_2.5 pollution days during 2013-2015 were divided into seven groups by cluster analysis at 100 m and six groups at 1000 m. The distributions of sea level pressure and geo-potential height, which were conducive to the accumulation of fine particles in the western Yangtze River Delta, were obtained by composite analysis. At 100 m, around 92% of trajectories of severe PM_2.5 pollution days were quite short, corresponding to slow moving weather systems. In vertical direction, those trajectories were below 950 hPa during the past 48 h, indicating that the transport mainly occurred in the near surface layer without evident upward or downward motions. Correspondingly, the sea-level pressure was homogeneous from North China to East China. At 1000 m, around 77% of severe pollution days belonged to short trajectory groups, while the broad region from Northwest China (Xinjiang) to Southeast China was controlled by high pressure systems at 850 hPa, and Anhui is located in the bottom of high pressure system or between two high systems. The results may be helpful for the forecast of PM_2.5 pollution.