Background: The COVID-19 pandemic indirectly impacts HIV epidemiology in Central/West Africa. We estimated the potential impact of COVID-19-related disruptions to HIV prevention/treatment services and sexual partnerships on HIV incidence and HIV-related deaths among key populations including female sex workers (FSW), their clients, men who have sex with men (MSM), and overall. Setting: Yaoundé (Cameroon) and Cotonou (Benin). Methods: We used mathematical models of HIV calibrated to city- and risk-population-specific demographic/behavioural/epidemic data. We estimated the relative change in 1-year HIV incidence and HIV-related deaths for various disruption scenarios of HIV prevention/treatment services and decreased casual/commercial partnerships, compared to a scenario without COVID-19. Results: A 50% reduction in condom use in all partnerships over 6 months would increase 1-year HIV incidence by 39%, 42%, 31% and 23% among MSM, FSW, clients, and overall in Yaoundé respectively, and 69%, 49% and 23% among FSW, clients and overall respectively in Cotonou. Combining a 6-month interruption of ART initiation and 50% reduction in HIV prevention/treatment use would increase HIV incidence by 50% and HIV-related deaths by 20%. This increase in HIV infections would be halved by a simultaneous 50% reduction in casual and commercial partnerships. Conclusions: Reductions in condom use following COVID-19 would increase infections among key populations disproportionately, particularly FSW in Cotonou, who need uninterrupted condom provision. Disruptions in HIV prevention/treatment services have the biggest impacts on HIV infections and deaths overall, only partially mitigated by equal reductions in casual/commercial sexual partnerships. Maintaining ART provision must be prioritised to minimise short-term excess HIV-related deaths.