Background: Although the role of clinical/biological factors associated with mortality has already been explored in HIV-infected patients on antiretroviral therapy (ART), to date little attention has been given to the potential role of social vulnerability. This study aimed to construct an appropriate measure of social vulnerability and to evaluate whether this measure is predictive of increased mortality risk in ART-treated patients followed up in the ANRS CO8 APROCO-COPILOTE cohort. Methods: The cohort enrolled 1,281 patients initiating a protease inhibitor-based regimen in 1997–1999. Clinical/laboratory data were collected every 4 months. Self-administered questionnaires collected psycho-social/behavioural characteristics at enrolment (month [M] 0), M4 and every 8–12 months thereafter. A multiple correspondence analysis using education, employment and housing indicators helped construct a composite indicator measuring social vulnerability. The outcome studied was all-cause deaths occurring after M4. The relationship between social vulnerability and mortality, after adjustment for other predictors, was studied using a shared-frailty Cox model, taking into account informative study dropout. Results: Over a median (IQR) follow-up of 7.9 (3.0–11.2) years, 121 deaths occurred among 1,057 eligible patients, corresponding to a mortality rate (95% CI) of 1.64 (1.37, 1.96)/100 person-years. Leading causes of death were non-AIDS defining cancers (n=26), AIDS (n=23) and cardiovascular diseases (n=12). Social vulnerability (HR [95% CI] =1.2 [1.0, 1.5]) was associated with increased mortality risk, after adjustment for other known behavioural and bio-medical predictors. Conclusions: Social vulnerability remains a major mortality predictor in ART-treated patients. A real need exists for innovative interventions targeting individuals cumulating several sources of social vulnerability, to ensure that social inequalities do not continue to lead to higher mortality.