OBJECTIVES We investigated why, despite its beneficial effect on the CD4 cell count, IL-2 therapy had no clinical benefit as shown in the ESPRIT and SILCAAT trials. We focused on subgroups of patients defined according to CD4 cell counts at baseline and over time to assess the threshold above which IL-2 therapy was no longer beneficial in a large cohort of HIV-1 infected patients. METHODS Within the French Hospital Database on HIV, a total of 953 IL-2-treated patients were compared with 27 750 IL-2-untreated patients, matched for the date of enrolment, sex, age, and the baseline CD4 cell count and plasma HIV-1 RNA level. The risk of clinical progression, defined as the occurrence of a new AIDS-defining event or death, was studied with multivariable Cox proportional hazards models and Poisson regression models. RESULTS We found no clinical benefit in patients starting IL-2 with CD4 count ≥200 cells/mm(3) [hazard ratio (HR) =1.13; 95% confidence interval (CI), 0.81-1.57], while a benefit was observed in patients with CD4 count <200 cells/mm(3) (HR=0.64; 95% CI, 0.48-0.86). The observed benefit was due to the risk reduction in the 100-350/mm(3) stratum of updated CD4 cell counts (relative rate=0.30; 95% CI, 0.09-1.03). CONCLUSIONS Higher CD4 cell counts at enrolment and shorter follow-up with low to intermediate CD4 cell counts may explain why IL-2 therapy had no observed clinical benefit in the SILCAAT study. Our findings suggest that the benefit of IL-2 is restricted to a narrow range of CD4 cell counts, arguing against the use of IL-2 in HIV infection to reduce the risk of clinical events.