Pires (2006) argues that Acc-ing and PRO-ing constructions belong to the same type of gerunds, namely clausal gerunds (CGs), and that the subject of CGs can be either a null subject (=PRO) or an overt DP with accusative Case. To account for this alternation, he adopts the mechanism of the movement theory of control. This paper argues that Pires’ (2006) analysis of CGs is problematic theoretically and empirically. Especially, the empirical data presented in this paper, which concern the ECM infinitival complement of consider, Subj-Aux inversion, the interpretation ambiguity of some PRO-ing gerunds, extraposition, and forming time adverbials, clearly suggest that Acc-ing and PRO-ing constructions behave differently in syntax. Based on these pieces of evidence, this paper concludes that the Pires’ (2006) movement analysis of PRO-ing constructions, suggested to integrate them with Acc-ing constructions under the name of CGs, should be abandoned. By extension, the movement analysis of the so-called TP-defective gerunds, suggested in order not to turn them counterexamples to Pires’ (2006) analysis, also should be abandoned. Then, their peculiar property of taking only the null subject is directly related to the predicates’ selection.