In the Knowledge Based Economy, research centres whether industrial or public, play a fundamental role. In terms of Knowledge Management, these organisations have a special status, because their production is knowledge and only knowledge. The Knowledge Capital they accumulate in their activities therefore is a strong strategic issue and the management of these assets has become crucial. The problem addressed in this paper is to design a pertinent methodology for Knowledge Management considering the specificity of knowledge production by research centres. This methodology is based on a suitable model to describe that knowledge production. The reference model is built on knowledge flows between the organisation and its knowledge workers, and a subsystem called “Knowledge Capital”. A research centre is defined by the fact that its product is only knowledge and is accumulated in its knowledge subsystem. Some economical characteristics of this Knowledge Capital are shown as being very adapted to knowledge produced in research centres. The methodology is based on two tools. The first tool is the knowledge map that can represent a comprehensive model of the Knowledge Capital of the organisation, which is often not well known or unstructured. That map is built on a shared and consensual vision of the main knowledge actors. It is not a map produced by a knowledge tool, but a co‑construction (through interviews) with the knowledge actors. The second tool is a grid for criticality analysis (Critical Knowledge Factors), which evaluates the knowledge domains of the organisation and suggests appropriate actions to be put in place for the most critical domains. This tool is a guide for interviewing knowledgeable actors in the organisation, to collect and analyse a set of data for decision support. The aim of the methodology is to provide a set of recommendations to build a KM plan of actions to preserve, share and make evolve the Knowledge Capital. The methodology has been elaborated through constant feed‑back with practice, and has been validated in many real cases in various countries. Three case studies (France, Brazil, and Canada) are succinctly described to exemplify the effectiveness of the methodology.