Viral RNA level in plasma is a sensitive experimental endpoint for evaluating the efficacy of AIDS vaccines or therapies in nonhuman primates. By quantifying viral RNA in the plasma of 77 rhesus monkeys for 10 weeks after inoculation with simian-human immunodeficiency virus 89.6P (SHIV-89.6P) or simian immunodeficiency virus mac 251 (SIVmac 251), we estimated variability in three viral load (VL) measures: peak VL, the postacute set point VL, and VL decline from peak. Such estimates of biological variability are essential for determining the number of animals needed per group and may be helpful for selecting the most appropriate measure to use as the experimental endpoint. Peak VL was positively correlated with set point VL for both viruses. Variability (standard deviation) was substantially higher in monkeys infected with SIVmac 251 than in those infected with SHIV-89.6P for set point VL and VL decline. The variability of peak VL was less than one-half that of set point VL variability and only about two-thirds of that of VL decline, implying that the same treatment-related difference in peak VL could be detected with fewer animals than set point VL or VL decline. Thus, differences in VL variability over the course of infection and between viruses need to be considered when designing studies using the nonhuman primate AIDS models.