Studies on disturbance events in riverine systems caused by environmental disasters and their effects on microbial diversity are scarce. Here, we evaluated the impact of the collapse of an iron ore dam holding approximately 50 million cubic meters of waste on both water and sediment microbiomes by deeply sequencing the 16S rRNA gene. Samples were taken from two impacted rivers and one reference river 7, 30 and 150 days postdisturbance. The impacted community structure changed greatly over spatiotemporal scales, being less diverse and more uneven, particularly on day 7 for the do Carmo River (the closest to the dam). However, the reference community structure remained similar between sampling events. Moreover, the impacted sediments were positively correlated with metals. The taxa abundance varied greatly over spatiotemporal scales, allowing for the identification of several potential bioindicators, e.g., Comamonadaceae, Novosphingobium, Sediminibacterium and Bacteriovorax. Our results showed that the impacted communities consisted mostly of Fe(II) oxidizers and Fe(III) reducers, aromatic compound degraders and predator bacteria. Network analysis showed a highly interconnected microbiome whose interactions switched from positive to negative or vice versa between the impacted and reference communities. This work revealed potential molecular signatures associated with the rivers heavily impacted by metals that might be useful sentinels for predicting riverine health.