Immunotherapy aims to assist the natural immune system in achieving control over viral infection. Various immunotherapy formats have been evaluated in either therapy-naive or therapy-experienced HIV-infected patients over the last 20 years. These formats included non-antigen specific strategies such as cytokines that stimulate immunity or suppress the viral replication, as well as antibodies that block negative regulatory pathways. A number of HIV-specific therapeutic vaccinations have also been proposed, using in vivo injection of inactivated virus, plasmid DNA encoding HIV antigens, or recombinant viral vectors containing HIV genes. A specific format of therapeutic vaccines consists of ex vivo loading of autologous dendritic cells with one of the above mentioned antigenic formats or mRNA encoding HIV antigens.This review provides an extensive overview of the background and rationale of these different therapeutic attempts and discusses the results of trials in the SIV macaque model and in patients. To date success has been limited, which could be explained by insufficient quality or strength of the induced immune responses, incomplete coverage of HIV variability and/or inappropriate immune activation, with ensuing increased susceptibility of target cells.Future attempts at therapeutic vaccination should ideally be performed under the protection of highly active antiretroviral drugs in patients with a recovered immune system. Risks for immune escape should be limited by a better coverage of the HIV variability, using either conserved or mosaic sequences. Appropriate molecular adjuvants should be included to enhance the quality and strength of the responses, without inducing inappropriate immune activation. Finally, to achieve a long-lasting effect on viral control (i.e. a "functional cure") it is likely that these immune interventions should be combined with anti-latency drugs and/or gene therapy.