Domestic Heating Ventilation and Air-Conditioning (HVAC) loads are among the most flexible residential loads for possible demand response (DR) applications. This paper aims at assessing the DR potential of HVAC loads in smart grids considering users' temperature preferences. For doing so, at first, a mathematical formulation is developed to evaluate the flexibility of HVAC loads. The model, by adjusting the HVAC load, intends to either maximize or minimize electricity consumption during specific hours of a day while the user thermal comfort is intact. The proposed model is then applied to a medium massive structure house in Helsinki, Finland. The study is performed for different seasons and the associated DR potential is evaluated. The simulation results showcase the significant potential of HVAC loads in both consumption reduction and load increment. It is shown that the DR potential can be significantly affected by temperature dead-bands representing consumers' willingness to accept change in their comfort. It is also demonstrated that installing even a small thermal storage capacity may lead to a greater value-added DR potential.