INTRODUCTION: Key populations, including female sex workers (FSW), people who inject drugs (PWID), and people in serodiscordant partnerships, experience higher HIV incidence compared to the general population. Maternal HIV retesting, particularly during late pregnancy, helps detect new infections and prevent vertical HIV transmission, but optimal testing schedules among key populations are unknown. METHODS: We used a Markov model to estimate the health and economic impacts of maternal HIV retesting on vertical HIV transmission outcomes among FSW and PWID in Kenya, South Africa, and pre-war Ukraine as well as among pregnant people in serodiscordant partnerships in Kenya and South Africa. We calculated incremental cost-effectiveness ratios (ICERs) for seven maternal retesting scenarios that included HIV testing during early antenatal care (ANC) and retesting from late ANC through nine months postpartum. RESULTS: Retesting during late ANC was estimated to avert 16% (Kenya), 14% (South Africa), and 8% (Ukraine) of infant HIV infections among key populations. Retesting during late ANC was cost-saving or cost-effective among all populations included in our model in South Africa and Kenya. In Ukraine, HIV retesting during late ANC was cost-saving for PWID but not cost-effective for FSW. Postpartum retesting was not cost-effective in any population. CONCLUSIONS: Maternal HIV retesting during late ANC is cost-saving or cost-effective for vertical HIV transmission outcomes for pregnant key populations and serodiscordant couples in Kenya and South Africa and is cost-saving for PWID in Ukraine.