The aim of this paper is to discuss the derivation of Tough-Constructions(TCs) on the basis of data from Spanish. Concretely, I argue that the movement analysis, according to which the matrix subject of TCs is derived by movement, cannot adequately account for the data. Hartman(2009) claims that the “defective intervention effect”(Chomsky, 2000) that TCs with a matrix experiencer display is a strong evidence for the movement. However, I claim that the intervention effect does not necessarily mean that the subject moved from the embedded clause to the matrix one. In support of my claim I discuss a set of Spanish data in which a clitic experiencer, an adverb or a PP experiencer appears in the matrix clause of TCs. I further claim that the movement analysis cannot account for why Spanish TCs do not tolerate split idiom chunks either. Finally, on the basis of the fact that Spanish TCs show neither the unboundedness nor wh-island effect nor the appearance of parasitic gaps, I suggest that the null operator movement within the infinitival embedded clause of TCs may not be universal but language-specific.