Because of their clinical and epidemiological consequences, persistent infections play an important role in shaping the selective pressures acting on host-microbe interactions. We can gain more about the contribution of persistent infections to the evolution of host-pathogen interactions by uncovering the dynamics of host defences. The present work takes a multi-angled approach to investigate the dynamics of host defences against persistent bacterial infections in Drosophila melanogaster. Firstly, in Chapter 1, by looking at the long-term dynamics of infection of various bacterial species, we investigated the conditions under which a pathogen persists or is cleared across various four bacterial species. All bacterial species could be cleared by the host, but the dynamics of clearance depended on the ability of the pathogen to exploit the host resources and reproduce. The most persistent bacteria (Lactococcus lactis and Providencia burhodogranariea) were those better at exploiting the host resources for their growth. Moreover, we could retrieve bacteria from hosts up to 78-days post-infection, marking an unprecedented estimation for the duration of persistence in insects. Besides virulence, based on the literature we had reason to believe that a previous exposure with a pathogen may enhance the ability of the host to limit or clear a persistent infection. In Chapter 2, we tested this hypothesis by using various methods to inactivate the pathogen for the primary encounter and exposing flies to L. lactis and P. burhodogranariea. Under the conditions tested, there was no advantage of a previous exposure in the face of a chronic infection. Hosts that are predicted to survive the infection while carrying a persistent infection show increased resistance in the early chronic phase, i.e., a lower bacterial load, compared to those predicted to succumb to an uncontrolled growth. To determine whether hosts predicted to have different infection outcomes vary in their tolerance, in Chapter 3, we measured fecundity-tolerance during a chronic infection with L. lactis and P. burhodogranariea. These two bacterial species caused a more pronounced decrease in fecundity over time in flies carrying a high load. However, only flies infected with L. lactis experienced a decrease in fecundity-tolerance, indicating that they are less able to counterbalance the fecundity costs associated to the infection. Chronically infected hosts sustain persistent antimicrobial peptide responses. In Chapter 4, we confirmed the presence of this response in hosts carrying a persistent P. burhodogranariea infection by measuring their protein expression in the chronic phase. In addition, we found that hosts may combine this antimicrobial response with nutritional immunity and a downregulation of other energetically costly branches of the immune system to fight the infection. The present work highlights the importance of considering infections as time- and context-dependent processes where both host and pathogen contribute to shape the outcome of infection.