The main aim of cocoa genetic improvement is to select clones or hybrid families with good agronomic, resistance and quality traits. In addition, the stability of the desired traits in time and space is also a very important selection criterion. This latter aspect has only been taken into account in a very limited way in the selection operations of the different producing countries. The trial network to be set up under the CFC/ICCO/IPGRI project provides a very good opportunity for obtaining reliable estimations of genotype x environment interactions. In order to do that, common experimental designs will need to be adopted and the data collected in the trials will have to be centralized for analyses. Procedures are proposed here for organizing data collection and their input into electronic datafiles. The possibilities for analyses are then discussed, aiming at optimum utilization of the data generated by the trial network.