Research evaluation of public institutes is important for measuring the performance of the science sector and improving the cost effectiveness and efficiency of public resource allocation. The purpose of this article is to construct a means of classifying publicly-funded R&D institutions and their performance (high or low). The main findings are two research performance functions, created using discriminant analysis with direct and Wilks methods, which have been successfully applied to 200 public research institutes belonging to the Italian National Research Council. The large number of indicators used within the models creates a macro index that produce, as output, the research performance score. The data analysed is from the 2001 period. The results show that 22.5% of public labs fall into the ‘high performance institute’ bracket and that they have a site larger than ‘low performance institutes’. The models are useful tools for decision making within the research bodies and for reducing X-inefficiency.