In individuals with HIV-1 infection, depletion of CD4+ T cells is often accompanied by a malfunction of CD8+ T cells that are persistently activated and/or exhausted. While the dynamics and correlates of CD4 counts have been well documented, the same does not apply to CD8 counts. Here, we examined the CD8 counts in a cohort of 497 Africans with primary HIV-1 infection evaluated in monthly to quarterly follow-up visits for up to 3 years in the absence of antiretroviral therapy. Statistical models revealed that (i) CD8 counts were relatively steady in the 3- to 36-month period of infection and similar between men and women; (ii) neither geography nor heterogeneity in the HIV-1 set-point viral load could account for the roughly 10-fold range of CD8 counts in the cohort (P > 0.25 in all tests); and (iii) factors independently associated with relatively high CD8 counts included demographics (age ≤ 40 years, adjusted P=0.010) and several human leukocyte antigen class I (HLA-I) alleles, including HLAA* 03:01 (P=0.013), B*15:10 (P=0.007), and B*58:02 (P < 0.001). Multiple sensitivity analyses provided supporting evidence for these novel relationships. Overall, these findings suggest that factors associated with the CD8 count have little overlap with those previously reported for other HIV-1-related outcome measures, including viral load, CD4 count, and CD4/CD8 ratio.