Tax evasion is reducing the revenues of public budgets of many European Union (EU) Member States (MS). To improve the effectiveness of tax collection, during the last decade authorities in several MS have taken measures to reduce the value added tax (VAT) gap (i.e., revenue received as a percentage of theoretical liability). In the Central and Eastern European region, VAT gap reduction measures have been implemented effectively in Hungary and Poland, whereas in Romania the effectiveness of these measures is very low: Romania has been the worst-performing EU MS in collecting VAT for more than 10 years. Our study analyses the factors influencing this VAT gap. Our analysis relied mainly on a fixed effects panel regression model, using for a balanced panel an individual and time fixed-effects with cluster-robust standard errors model, and for the unbalanced panel the fixed-effects regression with individual-specific slopes. Our results show that the size of the VAT gap is primarily influenced by five variables: the transparency index, the tax collection ratio, the law enforcement index, the VAT revenues ratio and the digitisation index.