Indonesia has the third highest number of people living with HIV/AIDS (PLWH) and the greatest increase in proportion of AIDS-related mortality in the Asia Pacific region between 2005 and 2013. Longitudinal mortality data among PLWH in Indonesia are limited. We conducted a retrospective cohort study from medical records of antiretroviral treatment (ART) recipients attending Badung General Hospital (BGH) and Bali Medica Clinic (BMC) between 2006 and 2014. We explored incidence of mortality by Kaplan–Meier analysis and identified predictors using a Cox proportional hazard model. In total, 575 patients were included in the analysis; the majority were male. The overall mortality rate was 10% per year. Multivariate analysis suggested that being male (adjusted hazard ratio [aHR]: 2.74; 95% confidence interval [CI]: 1.34–5.59), having a lower education (aHR: 2.17; 95%CI: 1.31–3.61), having heterosexual (aHR: 7.40; 95% CI: 2.61–21.00) or injecting drug use (aHR: 13.20; 95% CI: 3.17–55.00) as the likely transmission risk category, starting treatment with low CD4 cell counts (aHR: 3.18; 95% CI: 1.16–8.69), and not having a treatment supervisor (aHR: 4.02; 95% CI: 2.44–6.65) were independent predictors of mortality. The mortality was high, particularly in the first three months after initiating ART. These findings highlight the need to encourage HIV testing and early diagnosis and prompt treatment. Applying aspects of BMCs targeted HIV services model in more generalised services such as BGH may be beneficial. Providing adherence support as part of ART services is key to promoting adherence to ART.