Over the past 30 years, HIV has infected over 60 million people, with almost half succumbing to AIDS-related illnesses.While antiretroviral therapy, used to significantly reduce within-host HIV replication, was available within 10 years of the discovery of HIV/AIDS, it is only within the last 10 years that it has become truly effective and universally accessible. However, there are problems with this therapy, not least that it must be administered indefinitely , but is expensive and highly toxic. Furthermore, as therapy reaches more resource-limited regions, continual access can not be guaranteed, resulting in therapy interruptions. This, coupled with a significant cost reduction by systematically interrupting therapy, means a set of models which can account for both treatment events need to be developed, as numerous models exist for therapy introduction, but those for therapy removal are limited. Thus a set of delay differential models are designed, which account for previously overlooked important features of intracellular delay and HIV latency. Incorporation of these features requires additional model components, leading to a rapid increase in complexity. To combat this complexity issue, dimensional analysis is introduced, as a novel method of identifying key components to model function, thus allowing significant reduction in parameter space. Based on these developed models, a number of existing and potential treatment interruption regimes are investigated, with a best practice regime suggested.