The properties of excited nuclear matter and the quest for a phase transition which is expected to exist in this system are the subject of intensive investigations. High energy nuclear collisions between finite nuclei which lead to matter fragmentation are used to investigate these properties. The present report covers effective work done on the subject over the two last decades. The analysis of experimental data is confronted with two major problems, the setting up of thermodynamic equilibrium in a time-dependent fragmentation process and the finite size of nuclei. The present status concerning the first point is presented. Simple classical models of disordered systems are derived starting with the generic bond percolation approach. These lattice and cellular equilibrium models, like percolation approaches, describe successfully experimental fragment multiplicity distributions. They also show the properties of systems which undergo a thermodynamic phase transition. Physical observables which are devised to show the existence and to fix the order of critical behaviour are presented. Applications to the models are shown. Thermodynamic properties of finite systems undergoing critical behaviour are advantageously described in the framework of the microcanonical ensemble. Applications to the designed models and to experimental data are presented and analysed. Perspectives of further developments of the field are suggested.