Abstract: Analyzing and revealing the spatial and temporal characteristics of PM_2.5 in cities of the Yellow River Basin is of great significance for winning the battle of air pollution, promoting the establishment and improvement of the cross-regional cooperative control mechanism of air pollution, and the green high-quality development as well. Based on the PM_2.5 concentration monitoring data of 456 monitoring stations, this paper analyzed the spatial and temporal variation characteristics and evolution pattern of PM_2.5 in 70 cities in the Yellow River Basin from 2015 to 2021 by using the Moran index and standard deviation ellipse method, identified the pollution sources by Pearson correlation coefficient analysis method. The results showed that: (1) PM_2.5 concentration reflected monthly and seasonal changes. The monthly average concentration shows a ‘U-shaped’ distribution with a wide and slow bottom, reaching the maximum in December or January; in terms of seasonal, the average concentration in winter was the highest (1.9~2.6 times that in summer), followed by spring and autumn, and the lowest in summer, the annual average concentration decreased overall, showed a spatial differentiation of downstream > midstream > upstream (2) The spatial aggregation of PM_2.5 was characterized by ‘low-low’ agglomeration in the upstream, ‘high-high’ agglomeration in the downstream, and insignificant spatial aggregation in the midstream. The number of cities with positive spatial correlation agglomeration increased first and then decreased, while the number of cities with negative spatial correlation agglomeration is relatively small. (3) The spatial distribution pattern of PM_2.5 generally followed a ‘northwest-southeast’ geographical spatial trend,, the concentration distribution showed the characteristics of geographical spatial decentralization and gradual reduction of distribution range. (4) PM_2.5 pollution in upstream cities were complex and diverse, mainly including PM₁₀, NO₂, CO and SO₂, PM₁₀, NO₂ and CO were the main sources in midstream cities, while PM₁₀ and CO in downstream cities. The spatial heterogeneity of PM_2.5 concentration in cities of the Yellow River Basin is obvious, and there is a significant positive spatial correlation agglomeration in the upstream and downstream. Collaborative prevention and control in cities will help further improve air quality in the Yellow River Basin.