The detection of damage in structures using vibration data is one of the most studied subjects in the field of Structural Dynamics and Health Monitoring. This statement is easily verified when the amount of works concerning this theme is consulted. Usually the variation of mode shapes and natural frequencies after the occurrence of damage is used to estimate structural integrity. The central idea of this kind of methodology is to admit that damage mainly affects structural stiffness without significantly changes of the mass. This leads to small reductions of natural frequencies as well as slight alterations of mode shapes. However, during the modal identification procedure of a structure, there are some uncertainties inherent to this process due to temperature variation; noise embedded on data; the modal identification method applied; changes on structural loading; etc., that are able to induce small modal deviations that are not associated to damage. In that way, this paper simulates an undamaged instrumented structure in different scenarios of temperature and loading. Artificial noise is applied to the simulated data and the identified modal features of the tested structure are used to evaluate two well-known damage detection techniques. The cases analyzed suggest that the application of damage detection techniques must be carefully observed in order to avoid false-positive structural damage diagnosis.