Abstract In this short paper we introduce the notion of proof-based coverage. While traditional test coverage metrics measure which parts of the code are executed when a test passes, proof-based coverage aims to measure what parts of the code are required for a test to pass. Through formal verification, a proof of validity of an assertion is created. By mapping back the proof onto the source code, the essential parts can be marked, obtaining a novel coverage metric. We sketch a formalization of this notion, discuss the relationship to previous coverage metrics and show differences on small examples.