In this study, we analysed in a cohort of pregnant women followed for two years the proportion of women remaining at the same clinic, those who transferred to other clinics, and those lost to follow-up. The possible determinants of the loss to follow-up were also assessed in a setting of postpartum discontinuation based on CD4+ count. A total of 311 pregnant women received antiretroviral therapy from week 25 of gestational age until six months postpartum (end of breastfeeding period), or indefinitely if meeting the criteria for treatment (baseline CD4+ <350 cells/mm³). Twenty-four months after delivery, six women had died, 247 were in active follow-up, 21 had transferred to another antiretroviral therapy clinic and 37 were lost to follow-up (rate of loss to follow-up 13%, 95% CI 9.1–16.9%). The presence of a baseline CD4+ count above 350 cells/mm³ was associated with a ten-fold higher risk of loss to follow-up after six months of delivery (hazard ratio: 9.8, 95% CI 2.2–42.7, for baseline CD4 >350 cells/mm³ versus baseline CD4+ count below 350 cells/mm³, p = 0.002). This finding suggests that discontinuation of drugs when the risk of transmission has ceased can have a negative impact on the retention in care of these women.