Objective: The objective of this retrospective cohort study is to assess whether demographic and anthropometric parameters, laboratory tests, co-morbidity, co-infection, treatment regimen, IRIS and adherence predict response to HAART as measured by CD4 count, weight gain and functional status in a cohort of patients in Roma, the Kingdom of Lesotho. Method: Data were collected from a computerised database of the Antiretroviral Centre of the hospital. A cohort of 300 subjects was identified from hospital records from January 2007. Each of these subjects was followed up over a period of 12 months with data obtained for at least two visits within the 12-month span. Data were obtained on weight and CD4 at baseline, three months and also at six and 12 months, and data for haemoglobin were obtained only at 12 months. Variables that may be potential confounders were identified and univariate and multivariate logistic regression analyses were carried out to establish differences independent of confounding factors for the combined endpoints, as well as for each endpoint separately. Results: Three-hundred patient records were analysed. Approximately 70% of the patients had a CD4 increase of at least 150 cells over baseline values at the end of the review period and in 52.3% of the patients an increase in weight of 10% over baseline measurements was seen. Seventy-nine patients (26.3%) had a haemoglobin level of at least 14g/dL at 12 months, regardless of baseline values or gender. The inclusion of Zidovudine (AZT) in treatment regimens was found in 73% of the patients and in multivariate analysis AZT was associated with not having anaemia at the end of the review period. However there was a slight reduction in haemoglobin level in the first two to three months of therapy in comparison with both Stavudine (d4T) and Tenofovir (TDF) but not significant enough to result in clinical anaemia. Baseline CD4 values were similar for all treatments options but dissimilar in other outcome variables and continued to vary significantly throughout the review period. The outcomes of multivariate analyses suggest that the male gender appears to have better response to HAART as seen in each of the multivariate models. The most important determinant of haemoglobin response was baseline haemoglobin values. In the haemoglobin-associated multivariate model, HAART is associated with an increase in haemoglobin over baseline values. A history of TB prior to HAART was a major factor in weight response and it is thought to be as a result of IRIS, which is the unmasking of latent infections as the immune system reconstitutes. CD4 values have no direct influence on weight however, but an increase in weight was observed in all therapy groups. Conclusion: Clinical and immunological parameters can be used to monitor response to HAART and predict treatment outcomes. These parameters can be organised into monitoring tools that will be useful in resource-limited areas. This study suggests that AZT-containing regimens appear not to result in anaemia and that symptomatic anaemia might need additional investigation. Treatment with TDF appeared to have shown the best possible response pattern more but patients on TDF therapy will have to be included in the study to justify this observation. / Dissertation (MSc)--University of Pretoria, 2012. / Clinical Epidemiology / unrestricted