OBJECTIVE: To study retention in care (RIC) trajectories and associated factors in patients eligible for antiretroviral therapy (ART) in a universal test-and-treat setting (TasP trial, South Africa, 2012-2016). DESIGN: A cluster-randomized trial whereby individuals identified HIV positive after home-based testing were invited to initiate ART immediately (intervention) or following national guidelines (control). METHODS: Exiting care was defined as >/=3 months late for a clinic appointment, transferring elsewhere, or death. Group-based trajectory modeling was performed to estimate RIC trajectories over 18 months and associated factors in 777 ART-eligible patients. RESULTS: Four RIC trajectory groups were identified: (1) group 1 "remained" in care (reference, n = 554, 71.3%), (2) group 2 exited care then "returned" after [median (interquartile range)] 4 (3-9) months (n = 40, 5.2%), (3) group 3 "exited care rapidly" [after 4 (4-6) months, n = 98, 12.6%], and (4) group 4 "exited care later" [after 11 (9-13) months, n = 85, 10.9%]. Group 2 patients were less likely to have initiated ART within 1 month and more likely to be male, young (<29 years), without a regular partner, and to have a CD4 count >350 cells/mm. Group 3 patients were more likely to be women without social support, newly diagnosed, young, and less likely to have initiated ART within 1 month. Group 4 patients were more likely to be newly diagnosed and aged 39 years or younger. CONCLUSIONS: High CD4 counts at care initiation were not associated with a higher risk of exiting care. Prompt ART initiation and special support for young and newly diagnosed patients with HIV are needed to maximize RIC.< Réduire