ART regimens are challenging to follow, despite the simplicity of regimens of combined antiretroviral therapy (single pill once a day) and reduced toxicity. As near-perfect adherence is required, it is therefore important to gain better understanding of patients' perspectives on their ART. This thesis contains five observational studies in which we used data that were routinely collected in care, including long-term daily electronic adherence measurements and naturalistic observations of patient and provider statements, which are rare in the scientific literature. We also employed the novel machine learning method in the analysis. Our findings give insights about patient long-term ARV adherence (3-8 years, as per each study) and clinical outcomes and have potential for major impact in the field of ARV adherence by informing researchers and HCPs. Our findings also show that theory-based adherence enhancing interventions increase HIV patient retention in care and improve clinical and treatment outcomes.