While ethnography produces highly contextualized understandings of fields of practice, it is hard to assess whether the phenomena revealed through ethnographic study are typical on a regional level. To overcome this shortcoming, I introduce a research design, ethnographic upscaling, that combines in-depth ethnography with larger-N regional comparisons. While the in-depth ethnography provides valid hypotheses, comparison allows the testing of whether an observed phenomenon can be generalized. To illustrate ethnographic upscaling, I present research on the social engineering of water governance in Namibia. In the course of political “decentralization,” and inspired by community-based natural resource management policies, pastoralists have developed new rules stipulating how to share water and distribute the costs of providing it. While all communities were exposed to similar blueprints, local social practices concerning water governance differ. I show that these variations can be explained if we combine an ethnographic understanding of the dynamics within communities with an understanding of both networks that impinge on the individual cases and influences of external authorities.