The influence of previous structured treatment interruptions (STIs) on the length of time off therapy when highly active antiretroviral treatment (HAART) is discontinued in chronically HIV-infected subjects was assessed. A comparative, retrospective clinical cohort study included patients with plasma viral load (VL) <50 copies/ml and CD4 cell count >500 cells/mm(3) who interrupted HAART. Fifteen patients interrupted HAART after six 2-weeks-off-/4-weeks-on therapy cycles (STI group) and 30 subjects discontinued HAART without previous STIs (NSTI group). The criteria for treatment resumption were development of AIDS-defining clinical events, VL >100,000 copies/ml or CD4 <350 cells/mm(3). Median (IQR) time off therapy was 48 (29-56) weeks in the STI group and 31 (8-77) weeks in the NSTI group (p < 0.15). After 48 weeks, 46% of the patients in the STI group and 40% in the NSTI group remained off HAART (p < 0.74). No patient developed AIDS-defining events and all but one achieved virological control after treatment resumption. The CD4 nadir was 341 (298-464) cells/mm(3) among patients who reinitiated HAART and 560 (364-682) cells/mm(3) in those who remained off therapy by week 48 (p <0.01). Likewise, CD4 count prior to treatment interruption was 902 (806-1040) cells/mm(3) and 1123 (924-1234) cells/mm(3) in subjects resuming and remaining off HAART, respectively (p = 0.03). No relationship between treatment resumption and pre-ART VL or with the time with undetectable VL before enrollment was found. CD4 nadir was a significant predictor for treatment reinitiation in a multivariate analysis. Previous STIs do not influence time off therapy when HAART is definitively discontinued in chronically HIV-infected subjects. CD4 nadir is an important factor in the treatment discontinuation decision.