This paper will contrast the novel I-V test criteria with traditional and recent IDDq test methods. It compares their test effectiveness. We inted to show how I-V tests and IDDq tests fare in discriminating between "good" and "bad" dies and how test limits can be set empirically, especially for I-V testing. All results are based on data from an (internal) IBM experiment that was based on a large ASIC manufactured in a 0.18 ?m-Leff technology.