A variety of segregation measures have been proposed in the literature with the dissimilarity index and its variants being the most widely applied. While these measures provide a quick means of comparing segregation patterns across space and time, they are not linked to the processes that generate and maintain segregative patterns. This paper proposes a new method for measuring segregation. The method is based on comparative evaluations of neighbourhood characteristics as a determinant of spatial behaviour in cities. The proposed approach moves away from a purely geometric interpretation of segregation by anchoring the measurement of segregation into the actual urban setting. The method's main advantages are its responsiveness to changes in urban characteristics (e.g. spatial variations in rent increases), and changes in preference structures (e.g. a declining importance of distance as an impediment to relocation). A numerical example is used to demonstrate the behaviour of the proposed method against traditional segregation measures. The results indicate that the segregation experiences of minorities may be more severe than previously thought.