Code coverage considered harmful?

Summary
Measuring code coverage seems like a fool-proof way of evaluating the quality of your tests

Code coverage measurement is useful. But using it as a target is silly. It is too easy to increase coverage without adding any real value to the tests. Low code coverage in a certain component could indicate a problem with our test case design and as such is useful information. What you need to make sure is that people have access to the information and are given time to act upon it. But don’t ask them for minimum coverage.

• Nice overview of code coverage by the people of CCover
• Brian Marick on misusing code coverage
• Cem Kaner on Coverage

Other links:

• A Buyer’s Guide to Code Coverage Terminology
• Experience with the cost of different coverage goals for testing
• What is Wrong with Line Coverage