The Government Procurement Agreement (GPA) is one of the plurilateral trade agreements of the World Trade Organisation (WTO). It aims to liberalise trade in government procurement among its Parties. While it has many strong points, the GPA does not legally define what entities should actually covered by the Agreement. Notably, the evolution of the GPA has produced a list approach, whereby Parties list their ‘covered entities’ in a series of schedules. Unfortunately, the plurilateral nature of the GPA accession negotiation as well as the stringent reciprocity features of the list approach have complicated the accession negotiation, and discouraged Parties from providing a ‘wider’ rang of entity coverage. This turn of events has limited the effectiveness of the ‘Most-Favoured-Nation’ obligation, and thereby handicapped the GPA expansion. Moreover, this approach raises some tensions and lack of legal certainty as regards which are the entities covered by the GPA, especially in connection with entities that are not strictly ‘government entities’ such as State-owned enterprises (SOEs). This problem is exacerbated in the case of modern SOEs in developing countries, many of which can bear both public and private features. In order to avoid the drawbacks of the list approach mentioned above and response to the focal point in GPA accession negotiation, the author attempts to define ‘what is a government entity’ as a reference to address this problem raised by lack of precise definition on the entities that are subject to the GPA. Based on the analysis of the WTO adjudicatory reports under the multilateral agreements and the most recent influential free trade agreements, the thesis advanced is that generally, ‘commercial SOEs’ (namely SOEs engaging in commercial activities on a commercial basis), and ‘public SOEs’ (namely SOEs performing public functions under government controls or influence) must be treated differently under the GPA based on the nature of its activities, rather than their formal legal status. The thesis argues that only public SOEs should be covered, whereas commercial SOEs should not. The author arrived at this conclusion by conducting two parallel comparative studies, first, between public SOEs and ‘bodies governed by public law’, and second, between commercial SOEs with ‘public undertakings & undertakings with special or exclusive rights’, under EU Directive 2014/24/EU and Directive 2014/25/EU respectively. The studies clarified that a government entity must be defined by establishing an immediate and decisive causal link between government control and losing commercial freedom when carrying out activities. The existence of government control is identified from both the internal and external tier of the relationship between the entity and a government. Whether the entity competes with commercial freedom in the market where it carries out activities, can be judged by examination of the contestability of the market where the entity carries out its activities; the market power of the entity; and whether governments restrict the entrance to this market of new market competitors. Conclusively, the author submitted that the definition of ‘government entity’ consists of general scope and justification: generally, all governments and government-controlled entities shall be covered by the GPA at the first instance. However, if the controlled entity can prove that it operates in a contestable market (i.e. where it competes with other market players for commercial purpose), then the entity should be excluded from GPA coverage.