OBJECTIVE Our previous studies on CD4-guided therapy interruption (TI) showed that the durations of the first and second TIs were similar if antiretroviral therapy (ART) was resumed at a level of the CD4 cell count similar to or higher than the nadir CD4 T-cell count. Therefore, in a strategy of repeated CD4-guided TI, it is important to know which factors predict the time for the CD4 T-cell count to return to nadir (TRN). METHODS From a cohort of 125 patients who interrupted ART, 92 patients who reached a CD4 T-cell count similar to the nadir count were included in the study. RESULTS The median TRN was 12.3 months. In the multivariate analysis, younger age (P=0.011), lower pre-ART HIV RNA (P=0.022) and female gender (P=0.045) were associated with a longer TRN. After TI there were 11 clinical events in the group of patients whose nadir CD4 count was >200 cells/microL. Most of these events occurred when the TI was prolonged beyond the TRN. CONCLUSIONS The factors predicting the TRN were age, HIV RNA pre-ART and gender. Resumption of therapy at a CD4 cell count similar to the nadir CD4 count appears to protect against the development of clinical events. Given the observational nature of this study, no conclusions can be drawn regarding the possible application of TI in clinical practice.