The aim of this paper is to make recommendations for the procedural optimization of the Future Search Conference design on the basis of empirical evidence from two case study conferences in Germany and the United Kingdom. The paper presents the major criticisms that have been raised against the step-by-step conference design in the theoretical literature and contrasts these with the empirical findings of two stakeholder-based evaluations. The author draws attention to a number of weaknesses in the conference opening, the common ground phase and the action planning phase of the Future Search Conference design and makes proposals for design changes. The paper suggests that a systematic and stakeholder-oriented evaluation should be part of interventions like Future Search Conferences. The paper concludes with a reminder that the political context and local power relations are a key variable in determining success or failure of a Future Search conference. Copyright © 2002 John Wiley & Sons, Ltd.