Some APOBEC3 proteins cause G-to-A hypermutation in HIV-1 DNA when the accessory viral protein Vif is absent or non-functional. So far, cloning and sequencing has been performed to study G-to-A hypermutation. This is time-consuming and labour-intensive especially in the context of in vivo investigations where the number of hypermutated sequences can be very low. Thus, a massively parallel sequencing protocol has been developed for in-depth analysis of G-to-A hypermutation using the 454 pyrosequencing FLX system. Part of HIV-1 env was amplified and pyrosequenced after two rounds of infection in T cell lines and PBMCs using HIV-1 NL4-3Δvif. Specific criteria were applied to cope with major technical challenges: (1) the inclusion of hypermutated sequences, (2) the high genome diversity of HIV-1 env, and (3) the exclusion of sequences containing frameshift errors caused by pyrosequencing. In total, more than 140,000 sequences were obtained. 1.3-6.5% of guanines were mutated to adenine, most frequently in the GG dinucleotide context, the preferred deamination site of APOBEC3G. Non-G-to-A mutations occurred only in low frequencies (<0.6%). Single hypermutated sequences contained up to 24 G-to-A mutations. Overall, massively parallel sequencing is a very useful tool for in-depth analysis of G-to-A hypermutation in HIV-1 DNA induced by APOBEC3 proteins.