Over the course of evolution, retroviruses and their hosts have been in constant conflict, resulting in a long-term evolutionary "arms race" in which host and pathogen alternately gain selective advantages through functional innovations. On the host side, cells of higher eukaryotes have evolved to encode numerous inhibitory activities called restriction factors to block retroviral replication. Conversely, retroviruses have evolved to exploit numerous host cellular components to replicate and to efficiently counter host intrinsic lines of defence. The current acquired immunodeficiency syndrome (AIDS) pandemic exemplifies this struggle, as its causative agent the human immunodeficiency virus (HIV) was found to overcome several host-encoded restriction factors active against other retroviruses. TRIM5α is one such restriction factor, which inhibits retroviral infection soon after the virus enters its target cell by recognizing the capsid of incoming virions through its C-terminal PRYSPRY domain. Different TRIM5α alleles block distinct retroviruses in a species-specific manner: Human TRIM5α (huTRIM5α) potently blocks the so-called N-tropic murine leukemia virus (N-MLV), but is ineffective against the closely related B-tropic (B-MLV) and Moloney (Mo-MLV) strains, or against the simian immunodeficiency virus from macaque (SIVmac) and the two HIV strains (HIV1 and HIV2). In contrast, rhesus macaque TRIM5α (rhTRIM5α) potently blocks HIV1 and HIV2 but not N-, B-, Mo-MLV or SIVmac. My thesis aimed at gaining insight into the molecular mechanism of TRIM5α-mediated retroviral restriction. Firstly by investigating the huTRIM5α specific capture of retroviral capsids to better understand the ability of HIV1 to escape this recognition, and secondly by identifying cellular factors that interact with TRIM5α to subsequently determine their putative role in TRIM5α-mediated antiretroviral activity. In the first set of experiments, I demonstrated that substituting a single amino acid in the PRYSPRY domain of huTRIM5α expands its antiviral activity to B-MLV, Mo-MLV, SIVmac and HIV2, while HIV1 is still able to escape this block. Furthermore, I found that MLV could escape this restriction through the negative influence of four critical residues within its capsid, supporting a model whereby TRIM5α directly interacts with retroviral capsids. In the second part of my thesis, I exploited the remarkable conservation of the capsid among retroviruses to map the four key positions found in MLV onto the HIV1 capsid. I then confirmed that amino-acid substitutions at some of these positions could increase the sensitivity of HIV1 to huTRIM5α restriction. Further mutating an exposed cyclophilin A-binding loop of the HIV1 capsid could reinforce susceptibility to this factor. These results indicate that, despite major sequence divergences, HIV and MLV use structurally homologous residues to avoid capture by huTRIM5α. I then investigated the influence of the coiled-coil domain of huTRIM5α on the range of retroviruses recognised by this antiviral factor. I found that the PRYSPRY domain is necessary but not sufficient to dictate TRIM5α specificity as single amino acid changes within the coiled-coil region could modulate the range of targeted retroviruses. We favour a model whereby the coiled-coil domain exerts this effect indirectly, by influencing the orientation of the PRYSPRY domain and hence its ability to recognize retroviral capsids. Finally, I screened a human cDNA library for cellular proteins that interact with huTRIM5α. I found several hits, the physiological roles of which impact on cellular processes as diverse as ubiquitination and endocytosis. These results open possible avenues to pursue in future studies aimed at deciphering further the mechanisms by which TRIM5α blocks retroviruses.