Low birth weight, largely in the form of intrauterine growth retardation, has been used in animal studies as a sensitive indicator of adverse reproductive outcomes to suspect toxic agents. Methodological problems have severely curtailed studies of low birth weight for human risk assessment. For white and black births, we explore the use of statistical techniques to adjust for maternal risk factors and to isolate US counties having a significantly elevatea rate of unexplained low-birth-weight births in 1979. The data are derived from individual birth certificate information made available by the National Center for Health Statistics. Removing variation due to socioeconomic and other intrinsic factors available on birth certificates, clusters of high-risk counties appear. This paper discusses the methodology used to identify these counties.