Background We recently provided evidence that HIV-1 enters HeLa-derived TZM-bl and lymphoid CEMss cells by fusing with endosomes, whereas its fusion with the plasma membrane does not proceed beyond the lipid mixing step. The mechanism of restriction of HIV-1 fusion at the cell surface and/or the factors that aid the virus entry from endosomes remain unclear. Results We examined HIV-1 fusion with a panel of target cells lines and with primary CD4+ T cells. Kinetic measurements of fusion combined with time-resolved imaging of single viruses further reinforced the notion that HIV-1 enters the cells via endocytosis and fusion with endosomes. Furthermore, we attempted to deliberately redirect virus fusion to the plasma membrane, using two experimental strategies. First, the fusion reaction was synchronized by pre-incubating the viruses with cells at reduced temperature to allow CD4 and coreceptors engagement, but not the virus uptake or fusion. Subsequent shift to a physiological temperature triggered accelerated virus uptake followed by entry from endosomes, but did not permit fusion at the cell surface. Second, blocking HIV-1 endocytosis by a small-molecule dynamin inhibitor, dynasore, resulted in transfer of viral lipids to the plasma membrane without any detectable release of the viral content into the cytosol. We also found that a higher concentration of dynasore is required to block the HIV-endosome fusion compared to virus internalization. Conclusions Our results further support the notion that HIV-1 enters disparate cell types through fusion with endosomes. The block of HIV-1 fusion with the plasma membrane at a post-lipid mixing stage shows that this membrane is not conducive to fusion pore formation and/or enlargement. The ability of dynasore to interfere with the virus-endosome fusion suggests that dynamin could be involved in two distinct steps of HIV-1 entry - endocytosis and fusion within intracellular compartments.