In recent years, thanks to the increasing computational power available, real time computer animation has naturally evolved to model more complex and computationally expensive scenes. Consequently, all the problems concerning physical modelling need further research to tackle these new requirements, especially the problem of collision detection for deformable objects. Most existing solutions cannot not be trivially extended, because they are strongly based on the assumption that the shape of the object is fixed. In this paper we propose a general approach to reduce the cost of collision detection between deformable objects explicitly represented, regardless of the specific geometrical and physical manner in which they are modelled.