IntroductionThis article addresses a specific methodological challenge in social innovation research: how to select cases for multi-case analysis? In other words, it deals with the practical question of sampling 2 social innovation. Three connected reasons explain why this question deserves attention. First, knowledge on social innovation has been built to a large extent on case studies. This is understandable given the importance of contextual factors in innovation studies. Second, case study methodology clearly requires prior consideration of the type of cases that will be scrutinized. Third, and that point 1 This article is based on the results of the WISDOM research project (Innovation sociale dans l'accompagnement à domicile en Wallonie) funded by the Walloon Region (DG06, Germaine Tillion program, 2014Tillion program, -2017 and coordinated by the authors. They would like to thank their colleagues and their partners for their active contribution to the project, the anonymous reviewers for their thoughtful comments on the paper and express their gratitude to Dr. Anna Safuta for her careful proofreading. 2 Our sampling procedure differs from the logic referred to in probabilistic methods. Sampling here refers to the process of identifying (i.e. casing), sorting out and then selecting a certain number of units of analysis for further research through in-depth case studies.Creative Commons 4.0 International © Sinnergiak, Some Rights Reserved