A necessary step in the retroviral lifecycle is integration, the covalent insertion of the viral cDNA into the genome of the infected cell. This means that retroviruses, for example HIV, establish life-long infection. It also means that retroviruses are used as gene-delivery vectors to treat genetic diseases. Integration events are distributed non-randomly in the genome of the infected cell, with characteristic genus-specific preferences. This dissertation focuses on the lentiviral class of retroviruses, and explores two aspects of their integration: the means by which integration is targeted to its favored sites, and the consequences of integration at these sites for the host cell. The host protein LEDGF/p75 has been shown to interact with lentiviral integrases and contribute to their preference for integration in genes. We sought to establish the extent to which integration site selection is determined by LEDGF/p75 tethering. We first asked whether LEDGF/p75 was an essential integration tether, by analyzing integration site distribution in cells stringently depleted for LEDGF/p75. We found that LEDGF/p75 is responsible for much of the lentiviral integration preference, though probably not all. Secondly, we asked whether LEDGF/p75 tethering is sufficient to determine the genomic distribution of lentiviral integration. We used a fusion of LEDGF/p75’s integrase-binding domain and the heterochromatin-binding protein CBX1 to show that lentiviral integration could be retargeted away from its usual distribution and into CBX1-bound regions. These results underline LEDGF/p75’s central role in lentiviral integration, and the potential for manipulating its interaction with integrase. The effect of retroviral integration on the host cell is of particular relevance in gene therapy, where insertional activation of proto-oncogenes in patients is a serious concern. We present data on the genomic integration site distribution of a lentiviral vector for the correction of â-thalassemia in mice. While use of the same vector in a human patient led to clonal outgrowth, we report no evidence of insertional activation in the mouse model, but instead the suggestion that integration in genes may impart a growth disadvantage. This argues for the safety of lentiviral vectors, but raises questions about their effect on host gene expression.