We present a new hierarchical clustering criteria which can be appliedto data set. This is done after generating an initial partition byusinga Topological Self Organizing Map. This criteria contains two termswhich take into account two dierent errors simultaneously: the squareerror of the entire clustering (as the Ward criteria) and thetopological structure given by the Self Organizing Map. A parameter Tallows to control the corresponding in uence of these two terms. Resultson simulated data are presented which show the eect of this criteriafor different values of T.