A more recent paper that's related to this, "Coverage Is Not Strongly Correlated with Test Suite Effectiveness": http://www.linozemtseva.com/research/2014/icse/

A more recent paper that's related to this, "Coverage Is Not Strongly Correlated with Test Suite Effectiveness": http://www.linozemtseva.com/research/2014/icse/coverage/coverage_paper.pdf http://www.linozemtseva.com/research/2014/icse/coverage/cove... I wouldn't want people to conclude that coverage was a useless metric based on this, but it does seem to support the idea that chasing code coverage dogmatically is not necessarily the best use of your time.