Several recent public health crises have shown that the surveillance of zoonotic agents in wildlife is important to prevent pandemic risks. Rodents are intermediate hosts for numerous zoonotic bacteria. High-throughput sequencing (HTS) technologies are very useful for the detection and surveillance of zoonotic bacteria, but rigorous experimental processes are required for the use of these cheap and effective tools in such epidemiological contexts. In particular, HTS introduces biases into the raw dataset that might lead to incorrect interpretations. We describe here a procedure for cleaning data before estimating reliable biological parameters, such as bacterial positivity, prevalence and coinfection, by 16S rRNA amplicon sequencing on the MiSeq platform. This procedure, applied to 711 commensal rodents collected from 24 villages in Senegal, Africa, detected several emerging bacterial genera, some in high prevalence, while never before reported for West Africa. This study constitutes a step towards the use of HTS to improve our understanding of the risk of zoonotic disease transmission posed by wildlife, by providing a new strategy for the use of HTS platforms to monitor both bacterial diversity and infection dynamics in wildlife. In the future, this approach could be adapted for the monitoring of other microbes such as protists, fungi, and even viruses.