Characterization of optimal CTL epitopes in Gag can provide crucial information for evaluation of candidate vaccines in populations at the epicenter of the HIV-1 epidemic. We screened 38 individuals with recent subtype C HIV-1 infection using overlapping consensus C Gag peptides and hypothesized that unique HLA-restricting alleles in the southern African population would determine novel epitope identity. Seventy-four percent of individuals recognized at least one Gag peptide pool. Ten epitopic regions were identified across p17, p24, and p2p7p1p6, and greater than two-thirds of targeted regions were directed at: TGTEELRSLYNTVATLY (p17, 35%); GPKEPFRDYVDRFFKTLRAEQATQDV (p24, 19%); and RGGKLDKWEKIRLRPGGKKHYMLKHL (p17, 15%). After alignment of these epitopic regions with consensus M and a consensus subtype C sequence from the cohort, it was evident that the regions targeted were highly conserved. Fine epitope mapping revealed that five of nine identified optimal Gag epitopes were novel: HLVWASREL, LVWASRELERF, LYNTVATLY, PFRDYVDRFF, and TLRAEQATQD, and were restricted by unique HLA-Cw*08, HLA-A*30/B*57, HLA-A*29/B*44, and HLA-Cw*03 alleles, respectively. Notably, three of the mapped epitopes were restricted by more than one HLA allele. Although these epitopes were novel and restricted by unique HLA, they overlapped or were embedded within previously described CTL epitopes from subtype B HIV-1 infection. These data emphasize the promiscuous nature of epitope binding and support our hypothesis that HLA diversity between populations can shape fine epitope identity, but may not represent a constraint for universal recognition of Gag in highly conserved domains.