Abstract Article 107(1) of the Treaty on the Functioning of the European Union was originally designed to regulate the grant of subsidies by Member States, but it has been applied increasingly in recent years to regulate tax measures. Attempts to apply the State aid rules in this way have been met with controversy as they have been applied in unpredictable ways that have introduced significant changes to the existing tests in the case law for identifying State aid. These changes include a broader interpretation of the concept of selectivity, one of the criteria used to identify aid, which has been reoriented around the discrimination standard, whereby a measure is selective if it differentiates between categories of undertakings without being capable of justification by some legitimate public policy objective. Most controversially, this has facilitated the enforcement of the State aid rules against a wider range of State interventions. This article argues that the discrimination standard represents the most plausible and effective way of understanding selectivity among the alternative tests which have been proposed in the case law and academic literature. However, it goes on to identify a significant deficiency in the discrimination standard in the manner in which it is currently applied by the Union courts. The test inevitably requires certain value judgments to be made, but these are not made explicit by the existing case law. This article then proceeds to give further reflections on options for the refinement of the framework for assessing selectivity based on the discrimination standard and the existing logic of the case law and the objectives of the State aid control regime, which indicates how principled and coherent boundaries to the notion of State aid might be assessed in an individual case.