We have developed and evaluated a new method to quantify human immunodeficiency virus type 2 (HIV-2) proviral DNA based on LightCycler real-time PCR. The assay has a detection limit of 5 copies/10(5) peripheral blood mononuclear cells (PBMC) and is insensitive to HIV-2 strain variability: HIV-2 subtypes A and B are both recognized and quantified. The intra- and interassay coefficients of variation range from 16 to 40% for high provirus concentrations (5 x 10(5) copies) and from 41 to 39% for low concentrations (5 copies). We used this method to compare the proviral DNA load and viral RNA load in plasma with clinical and immunological status for 29 patients infected by HIV-2 (subtype A in 17 and subtype B in 12). The proviral load (median, 201 copies/10(5) PBMC) was similar to that reported for HIV-1 infection. The median proviral loads did not correlate with the CD4(+) cell count categories and were as follows for CD4(+) cell counts of >400, 200 to 400, and <200 cells/mm(3), respectively: 121 copies/10(5) PBMC (n = 8; range, <5 to 712 copies/10(5) PBMC); 114 copies/10(5) PBMC (n = 9; range, <5 to 1,907 copies/10(5) PBMC); and 285 copies/10(5) PBMC (n = 12; range, 53 to 2,524 copies/10(5) PBMC). Proviral load did not correlate with plasma HIV-2 RNA positivity. As HIV-2 is considered to replicate less efficiently than HIV-1, these high proviral loads might be explained by the proliferation of infected cells.