Records from existing climate networks are burdened with significant data inhomogeneities due to station moves, instrument changes, and changes in observation procedures, often of undetermined date or character. These inhomogeneities make it very difficult to separate true climate trends from other sources of climate variability. The U.S. Climate Reference Network (US CRN) was established provide well-calibrated and well-characterized baseline information in support of research into climatic change and variability across the United States. The network will be deployed in phases. The initial distribution will provide more or less geometrically uniform coverage across the contiguous United States. Location of the first fifty stations is guided by the need to capture the annual average climatic signal for the contiguous US. An important constraint on site locations is the need for pieces of property with stable ownership and usage practices remaining essentially unchanged for many decades to come, to provide environmental stability. This paper will summarize experiences in attempting to locate places that meet the criteria necessary to accomplish the desired goals of the network. Long-term stable environments are taken here to mean essentially uninfluenced for 50 years or more by significant changes to the immediate environmental surroundings. This is crucial with regard to possible future encroachments by human structures. Sites are assessed on their suitability to detect, monitor, and quantify climatic trends and variations unduly influenced by local environmental factors like topography, proximity to a body of water etc. Stations will be located to ensure that major nodes of the Nation's climate variability are captured while accounting for regional spatially representativeness, including orographic, biotic, and other environmental factors. High-risk sites are avoided; such as flood plains or low areas adjacent to river basins, estuaries, and coastal offshore barrier islands); nearby rock walls/cliffs and other blockages; and persistent periods of extreme snow depths (e.g., several meters/tens of feet). As a practical matter sites must initially be located where electric power can be accessed. The eventual selection of a US CRN instrument site will be the result of a balance between competing demands, such as those highlighted above and an assessment of the "quality of measurements." We employ a classification scheme described by Leroy (1998) to document the "meteorological measurements representativity" of each site, and discuss issues arising with trying to employ that scheme in the desired geographical settings.