The invention concerns a method to evaluate whether a statistical time delay (TD) between a first event and a second event of a device under test is better than a test limit (TL). The method includes the steps: performing a minimum number N of tests and evaluating the time delay (TD) from each test; modeling a first probability distribution (P1) of the evaluated time delays (TD); obtaining a second probability distribution (P2) of the evaluated time delays (TD); performing a statistical transformation in order to obtain a third probability distribution (P3) of the evaluated time delays (TD); and deciding to pass the device under test, if a certain percentage of the area of the third probability distribution (P3) is on a good side (GS) of the test limit (TL2).