In order to design strategies for eradication of HIV-1 from infected individuals, detailed insight into the HIV-1 reservoirs that persist in patients on suppressive antiretroviral therapy (ART) is required. In this regard, most studies have focused on integrated (proviral) HIV-1 DNA forms in cells circulating in blood. However, the majority of proviral DNA is replication-defective and archival, and as such, has limited ability to reveal the dynamics of the viral population that persists in patients on suppressive ART. In contrast, extrachromosomal (episomal) viral DNA is labile and as a consequence is a better surrogate for recent infection events and is able to inform on the extent to which residual replication contributes to viral reservoir maintenance. To gain insight into the diversity and compartmentalization of HIV-1 under suppressive ART, we extensively analyzed longitudinal peripheral blood mononuclear cells (PBMC) samples by deep sequencing of episomal and integrated HIV-1 DNA from patients undergoing raltegravir intensification. Reverse-transcriptase genes selectively amplified from episomal and proviral HIV-1 DNA were analyzed by deep sequencing 0, 2, 4, 12, 24 and 48 weeks after raltegravir intensification. We used maximum likelihood phylogenies and statistical tests (AMOVA and Slatkin-Maddison (SM)) in order to determine molecular compartmentalization. We observed low molecular variance (mean variability ≤0.042). Although phylogenies showed that both DNA forms were intermingled within the phylogenetic tree, we found a statistically significant compartmentalization between episomal and proviral DNA samples (P&10 -6 AMOVA test; P = 0.001 SM test), suggesting that they belong to different viral populations. In addition, longitudinal analysis of episomal and proviral DNA by phylogeny and AMOVA showed signs of non-chronological temporal compartmentalization (all comparisons P&10 -6) suggesting that episomal and proviral DNA forms originated from different anatomical compartments. Collectively, this suggests the presence of a chronic viral reservoir in which there is stochastic release of infectious virus and in which there are limited rounds of de novo infection. This could be explained by the existence of different reservoirs with unique pharmacological accessibility properties, which will require strategies that improve drug penetration/retention within these reservoirs in order to minimise maintenance of the viral reservoir by de novo infection. © 2011 Buzón et al.