Jails are often referred to as "reservoirs of disease" and presented as the origin of infectious diseases or for the development of chronic diseases. The present article argues that the data behind this metaphor are generally taken from nonrepresentative studies by analyzing non-TB-focused studies of jail inmate health issues published in both health and criminal-justice-related journals. Issues such as the use of extremely large jail systems, geographic location, diseases studied, and the lack of attention to jail processes are examined. The article explores the "length of stay" variable from booking to release known as "flow" and how the flow of detainees through jails over time affects who is left for data collection. Data on releases from a large Florida urban jail system for a 1-year period are analyzed to determine the impact of factors associated with release decisions. These, in turn, are related to the representativeness and generalizability of the observed data in previous research and analyzed in terms of potential bias for our understanding of the relationship between jail populations and community health. Issues of health-focused practice standards and recommendations issued without reference to "flow" factors are discussed. Finally, suggestions how criminologists and criminal justice researchers can inform the research on health among jail inmates are canvassed. (Published Abstract)