Construction defects, even visually untraceable, can forfeit the acoustic performance of systems. The execution of the systems is a parameter that is hard to control. This article aims to evaluate construction defects. Wall system samples of real proportions were made with induced defects. In comparison with the reference, reductions of 1 dB were found for the samples with defective gaps above the wall and 2 dB for the trials with no vertical filling. Regarding the samples with installations, this reduction reached 1 and 3 dB for samples with installation of electrical junction boxes at the same face and two on each face, respectively.