BACKGROUND: Although Uganda rolled out Differentiated Service Delivery(DSD) models in June 2017 to improve retention and viral load suppression rates among clients on Highly Active Antiretroviral Therapy (HAART), these have remained low relative to the Joint United Nations Programme on HIV/AIDS(UNAIDs) targets of achieving 95% population with HIV tested, 95% of tested positive clients for HIV to be on Highly active Antiretroviral therapy and 95% of clients On Antiretroviral therapy be suppressing by 2030(95-95-95 UNAIDS targets). The purpose of this study was to determine sustained retention, viral load suppression and their determinants among clients on HAART enrolled under different Differentiated service delivery models in Katakwi district in Eastern Uganda. METHODS: A retrospective cohort study of clients enrolled on HAART in the different approaches of DSD who were active by 2017 and followed up to 2020 was done. The primary outcomes included sustained retention, viral load suppression and their determinants among clients HAART in different DSD approaches. Eight health facilities providing HAART services were purposively sampled and 771 clients on HAART were sampled out by simple random selection from a total population of 4742 clients on HAART in Katakwi district. We analysed retention, viral load suppression rates, and their determinants by logistic regression method using STATA. RESULTS: A total of 771 participants were sampled of whom 42.7% were male and 57.3% were female, with the mean age being 40 years. Retention rates at 95% CI of participants were 99.35% at 12 months, 94.03 at 24 months, 89.88% at 36 months and 84.57% at 48 months. The viral load suppression rates were 57.3% at 12 months, 70.3% at 24 months, 70.3% at 36 months and 69% at 48 months. Retention was higher in the community based DSD model as compared to the facility-based model. Viral load suppression was higher in the community based DSD models in which Community Drug Distribution Points had the highest achievement (92%) followed by Community Client-Led ART Distribution (79%) compared to the facility based DSD models in which Facility Based Individual Management performance (34.3%) was far below the set standard of 95%, followed by Facility Based Groups (65%) with Fast Track Drug Refill having relatively better performance (80.9%). Being 40-59 years, receiving care from the general hospital, being married, having good current adherence, being on the first line of the current regime and being a female are other predictors of viral load suppression, whereas being 40-59 years of age, having good current adherence, being on the current first-line regime and having no co-morbidities were predictors of good retention. CONCLUSIONS: generally, facility and community based DSD models have demonstrated improved retention and viral load suppression. However, community-based models have shown to be more effective than facility-based models through mitigation of barriers to effective HIV/AIDS care of clients on HAART. Viral load suppression remained below the UNAIDs target of 95% by 2030, albeit it improved over time.