Depth of Inheritance Tree (DIT) is supposed to be a factor influencing the cost of testing: test would be more expensive if DIT is large. A question is thus to know whether DIT can be used as a predictive metric to estimate the cost of testing. In this paper, we consider the cost of testing as the number of test cases required to achieve the branch coverage, which is a classical criterion for structural testing, and which is given by McCabe's Cyclomatic Complexity metric. We analyzed 25 applications to identify if the DIT is good test cost indicator. This paper shows that DITA is too abstract to be really relevant to predict the cost of testing.