We consider a viral genome Dabrafenib length of L = 100 nucleotides, which spans the experimental fitness landscape and captures expected epistatic interactions between multiple loci and is also similar to the genome lengths examined in the experiments we consider. We assume that mutations occur independently at each of the L sites at the rate m =361025 substitutions per site per replication. Because a majority of HIV mutations are transitions, we ignore transversions, insertions and deletions. Following recent estimates, we choose the recombination rate r =8.361024 crossovers per site per replication. The relative fitness of the founder strain, determined by j, remains unknown. Recent studies show that the founder strain evolves under selective forces and is distinct from the strain dominant in the chronic infection phase. In a previous study, we found that j,0.05�C0.1 provides best fits to the evolution of divergence and diversity in the two patients examined. This is in accordance with the maximum divergence of,0.1 in the patient data we consider. Here, we therefore set j = 0.1. We let selection follow the fitness landscape determined experimentally. We also examine the effects of alternative fitness landscapes on our estimates of Ne. The frequency of multiple infections of cells in vivo remains uncertain. Infections of individual cells by multiple virions allow the formation of heterozygous progeny virions and set the stage for recombination to introduce genomic variation. Jung et al. found that infected splenocytes in the two patients they examined harbored between 1 and 8 proviruses with a mean of 3�C 4 proviruses per cell. In contrast, Josefsson et al. recently observed that a vast majority of the peripheral blood mononuclear cells in four patients harbored single proviruses. Here, we therefore perform simulations with both these patterns of multiple infections of cells: We first follow Jung et al. and let each cell be infected by M=3 virions. We then repeat our simulations with M drawn from a distribution that follows from a model of viral dynamics and that mimics the observations of Josefsson et al.. Although the viral burst size is large,,102�C104 only a few of the virions produced may be infectious. More recent estimates of the basic reproductive ratio of HIV-1 in vivo suggest the MLN4924 company production of 6�C8 infectious virions per cell. Here, we let each cell produce P=5 infectious progeny virions. We let simulations proceed to 4000 generations. This influence of recombination is consistent with current population genetics theories : When C is small, random genetic drift creates negative linkage disequilibrium according to the Hill-Robertson effect.