Cyanobacterial blooms, which not only threaten the health and stability of aquatic ecosystems but also influence the microbial community within, emerges as one of the most concerning problems in China. However, how cyanobacterial blooms affect the spatiotemporal variation of aquatic microbial communities remains relatively unclear. In this study, we used high-throughput sequencing to investigate how the cyanobacterial and bacterial community spatiotemporally vary along with main cyanobacterial bloom phases in upstream rivers of a eutrophicated water source reservoir. Both cyanobacterial and bacterial diversities in each river were significantly lower (P < 0.05) during the bloom outbreak phase, showing the apparent influence of cyanobacterial bloom. Dominant cyanobacterial taxa included Cyanobacteriales and Synechococcales, and dominant bacterial taxa comprised Acinetobacter, CL500-29, hgcI clade, Limnohabitans, Flavobacterium, Rhodoluna, Porphyrobacter, Rhodobacter, Pseudomonas, and Rhizobiales, whose changes of relative abundance along with the bloom indicated distinct community composition. Non-metric multidimensional scaling analysis proved that community composition had significant difference amongst bloom phases. Linear discriminant analysis (LDA) with LDA effect size analysis (LEfSe) identified unique dominant cyanobacterial and bacterial OTUs at different phases in each river, indicating spatiotemporal variations of communities. Canonical correlation analysis or redundancy analysis revealed that at different bloom phases communities of each river had distinct correlation patterns with the environmental parameters (temperature, ammonium, nitrate, and total phosphorus etc.), implying the spatial variations of microbial communities. Overall, these results expand current understanding on the spatiotemporal variations of microbial communities due to cyanobacterial blooms. Microbial interactions during the bloom may shed light on controlling cyanobacterial blooms in the similar aquatic ecosystems.