HIV is notorious for its ability to evade immunity and anti-viral drugs through rapid evolution. Thus, efforts to create an HIV vaccine aim to target conserved parts of the virus likely to have a limited evolutionary capacity for escape. However, our understanding of HIV’s evolutionary capacity is incomplete. To further this understanding, we used deep mutational scanning to experimentally estimate the effects of all ≈104 possible single amino-acid mutations to most of HIV’s envelope protein (Env) on viral replication in cell culture. First, we made a library of env genes with ~1-2 codon-level mutations per gene and with nearly all possible single amino-acid mutations. Next, we selected for functional variants that supported viral replication in human T-cells. We then used next-generation sequencing to measure the amount that each mutation was enriched or depleted upon selection. From this data, we estimated each site’s preference for each of the 20 amino acids using a statistical framework and software package developed by our laboratory. We compared our estimates of Env’s site-specific amino-acid preferences in cell culture to the actual frequencies of these amino acids in naturally occurring HIV sequences. Our measurements are strongly correlated with amino-acid frequencies in natural sequences for most sites known to be important for conserved protein functions such as receptor binding. While this correlation is also high for most buried sites, it is dramatically lower for surface-exposed sites that are subject to pressures not present in our experiments such as antibody selection. We plan to further explore the mutational tolerance at antibody epitopes to identify possible antibody-escape mutations. Ultimately, we plan to repeat this experiment in the presence of antibody selection to identify actual escape mutations. Overall, our estimates of Env’s mutational tolerance in cell culture provide a basis to better understand HIV’s evolutionary capacity in nature.