Numbers are the primary way that we know about AIDS in Africa, yet their power and utility often obscure the conditions of their production. I show that quantification is very much a sociocultural process by focusing on everyday realities of making AIDS-related numbers in Malawi. "Seeing like a research project" implies systematically transforming social reality into data points and managing uncertainties inherent in numbers. Drawing on 20 months of participant observation with survey research projects (2005, 2007-2008), I demonstrate how standards govern data collection to protect and reproduce demographers' shared expectations of "high-quality data." Data are expected to be "clean," accurate and precise, data collection efficient and timely, and data collected from sufficiently large, pure, and representative samples. I employ ethnographic analysis to show that each of these expectations not only guides survey research fieldwork but also produces categories, identities, and practices that reinforce and challenge these standardizing values.