Griffiths method is a widely used to estimate the comfort temperature of occupants. Although 0.5/K has been widely used as the representative thermal sensitivity (Griffiths constant) on various building types, this value was derived mainly from office data relying on assumptions that have not been fully field-validated (i.e., the assumption of no adaptive behavior occurring during the day). With an aim to empirically derive the thermal sensitivity (Griffiths constant) of occupants in residential settings, field monitoring was conducted in eleven households in South Korea. Within each household, physical indoor environmental data, subjective data and occupant behavioral data were recorded at 5 min intervals (from 8 a.m. to 8 p.m.). The value of the resulting regression coefficient (b) between the indoor temperature and the thermal sensation of occupants was estimated to be 0.356/K. This result indicates that the thermal sensation of the occupants of residential buildings is less sensitive to indoor temperature changes compared to that of occupants of other building types (e.g. offices 0.5/K). To increase the reliability of future thermal comfort research, the thermal sensitivity should reflect the characteristics of the building in question, rather than applying uncritically a generalized, universal Griffiths Constant such as 0.5/K.