Fish community structure could be directly and indirectly affected by biotic, abiotic and anthropogenic factors. Although fish community suffered less disturbance than big rivers, moderate disturbance could also change fish community structure in mountain rivers due to environmental variables such as current velocity, water depth, water temperature, dissolved oxygen and acidity. However, there are limited information about changes in fish composition in mountain river where less human disturbance. So the objective of this study was to elucidate fish diversity and community assembly patterns and identify relationships between environmental parameters and fish community structures. This survey was conducted seasonally at 16 locations in five rivers in Pingxiang city from June 2019 to May 2020. A total of 56 species were collected, and 16 fish species were endemic to China. The index of relative importance showed that Rhodeus ocellatus was the most dominant species in these rivers. The fish species showed low abundance and diversity in Pingxiang city. Cluster and non-metric multidimensional sequencing analyses showed clear spatial segregation, which divided the 16 sample sites into four significantly different groups. Permutational analysis of variance showed a significant difference in fish composition among the sites (p = .002). Furthermore, similarity percentages analysis showed that 15 fish species contributed to the greatest extent, and 13 fish species were divergent among the groups. The redundancy analysis revealed that water depth, pH and water temperature were the predominant factors contributing to the variation in fish composition. Thus, these results could guide the restoration and spatial management of fish resources as well as provide theories for the formulation of relevant conservation policies.