The authors propose a new method for early detection of disease outbreaks using only case numbers, without requiring population-at-risk data. This space-time permutation scan statistic is designed to minimize assumptions about the outbreak's characteristics and adjust for natural spatial and temporal variations. The method was evaluated using daily analyses of hospital emergency department visits in New York City, detecting four strong signals that likely preceded citywide outbreaks of rotavirus, norovirus, and influenza. The number of false signals was minimal. The method is practical for local and national health departments setting up early disease detection surveillance systems, especially in developing countries where population-at-risk data may be unavailable or unreliable. The authors acknowledge the importance of data quality checks and adjustments for missing data, and note that the method is sensitive to the spatiotemporal characteristics of the outbreak.The authors propose a new method for early detection of disease outbreaks using only case numbers, without requiring population-at-risk data. This space-time permutation scan statistic is designed to minimize assumptions about the outbreak's characteristics and adjust for natural spatial and temporal variations. The method was evaluated using daily analyses of hospital emergency department visits in New York City, detecting four strong signals that likely preceded citywide outbreaks of rotavirus, norovirus, and influenza. The number of false signals was minimal. The method is practical for local and national health departments setting up early disease detection surveillance systems, especially in developing countries where population-at-risk data may be unavailable or unreliable. The authors acknowledge the importance of data quality checks and adjustments for missing data, and note that the method is sensitive to the spatiotemporal characteristics of the outbreak.