In this presentation, we will talk about the results of an analysis of doi’s from real life systems - with emphasis on invalid doi’s. The analysis is based on a representative sample from approximately 500.000 doi's from app. 80 different institutions (research institution and funders). The non-matched doi’s are categorised as 1) valid doi’s not covered by Dimensions, 2) valid doi’s that resolve as a dead link and 3) invalid doi’s that are non-resolvable. The analysis is ongoing and hence categorisation and sub-categorisation is in development. DOI is one of the best adopted PID we have - and hence a PID which can trigger interesting discussion points based on real-life usage of PID’s