Machine Translation (MT) systems deliver translations instantly; however, users can easily identify MT output given that it may contain inaccurate word combinations, literal incorrect and ambiguous translations. For mitigating the inaccuracies associated with MT, the dissertation explores the feasibility of training a Customised Machine Translation (CMT) system in the context of Languages for Specific Purposes (LSP) for the legal domain. KantanMT a cloud-based CMT platform was deployed to translate English-Arabic UN Resolutions. UN Resolutions, a type of LSP, have specific characteristics including standardized style, fixed expressions and specialized terminology. The study adopts empirical and qualitative approaches to analyze and evaluate the translation produced by CMT in terms of the... overall translation quality and recurring linguistic errors. The analysis of the study concludes that LSP is highly feasible with CMT systems as shown by the highly accurate automatic and manual evaluations. Furthermore, the linguistic analysis of LSP can be reused as a reference for training CMT, particularly in the context of Statistical Machine Translation (SMT), as well as performing manual evaluation and adopting pre-editing and post-editing strategies.