The prevent mother-to-child transmission (PMTCT) "cascade" describes the programmatic steps for pregnant and breastfeeding women that influence HIV transmission rates. To this end, HIV-infected pregnant women and mothers need access to health services and adhere to antiretroviral (ARV) prophylaxis or lifetime treatment. Within the cascade, the concept of "retention-in-care" is commonly used as a proxy for adherence to ARV interventions and, even, viral suppression. Yet surprisingly, there is no standard definition of retention-in-care either for the purposes of HIV surveillance or implementation research. Implicit to the concept of retention-in-care is the sense of continuity and receipt of care at relevant time points. In the context of PMTCT, the main challenge for surveillance and implementation research is to estimate effective coverage of ARV interventions over a prolonged period of time. These data are used to inform program management and also to estimate postnatal MTCT rates. Attendance of HIV-infected mothers at clinic at 12-month postpartum is often equated with full retention in PMTCT programs over this period. Yet, measurement approaches that fail to register missed visits, or inconsistent attendance or other missing data in the interval period, fail to capture patterns of attendance and care received by mothers and children and risk introducing systematic errors and bias. More importantly, providing only an aggregated rate of attendance as a proxy for retention-in-care fails to identify specific gaps in health services where interventions to improve retention along the PMTCT cascade are most needed. In this article, we discuss how data on retention-in-care can be understood and analyzed, and what are the implications and opportunities for programs and implementation research.