In the extensive literature on political parties, the ethnographic method appears to be scarcely used. Since the pioneering work by Michels, which to some extent relied (albeit implicitly) on “participant observation”, this approach suffered a long eclipse in the English-language literature. Nowadays, international research favors the confrontation of big datasets, whether relating to membership, leadership, organizational reforms, or programmatic contents. The recent blossoming of a “political ethnography” (Auyero, Joseph and Mahler 2007) in sociology and anthropology departments in the United States has mainly benefited analysis of social movements. Applied to parties, it is ultimately in French works that the ethnographic approach has been most used; indeed, it constitutes a strong marker of the gap between the French and English-language political science. Originally prompted by the prominent role of the French Communist Party often described as a genuine “counter-society” (Kriegel, 1970), this approach has been spreading for a few decades on the study of membership and leadership within mainstream parties. This paper, relying on a broad literature review and on results drawed from the research led by the author on the French “Parti socialiste”, aims to provide an insight on this “French tradition” (whether applied to French organizations, or to parties from other countries). It addresses both of its methodological difficulties and analytical benefits.The first part reviews the empirical characteristics of parties that may complicate and, simultaneously, give easier access to fieldwork, and the second part explores the specificities of a research relationship that is necessarily “political”. The third part deals with the theoretical contributions of ethnographic work to major issues such as the analysis of parties as socializing institutions and the study of factionalism.