In July 1995, LANL hosted an accident dosimetry intercomparison. When all reactors on the Oak Ridge Reservation were idled in 1988, the Health Physics Research Reactor (HPRR), which had been used for 22 previous intercomparisons dating from 1965, was shut down for an indefinite period. The LANL group began characterization of two critical assemblies for dosimetry purposes. As a result, NAD-23 was conceived and 10 DOE facilities accepted invitations to participate in the intercomparison. This report is a summary of the performance of one of the participants, the Centralized External Dosimetry System (CEDS). The CEDS is a cooperative personnel dosimetry arrangement between three DOE sites in Oak Ridge, Tennessee. Many successes and failures are reported herein. Generally, the TL dosimeters performed poorly and always over-reported the delivered dose. The TLD processing procedures contain efforts that would lead to large biases in the reported absorbed dose, and omit several key steps in the TLD reading process. The supralinear behavior of lithium fluoride (LiF) has not been characterized for this particular dosimeter and application (i.e., in high-dose mixed neutron/gamma fields). The use of TLD materials may also be precluded given the limitations of the LiF material itself, the TLD reading system, and the upper dose level to which accident dosimetry systems are required to perform as set forth in DOE regulations. The indium foil results confirm the expected inability of that material to predict the magnitude of the wearer's dose reliably, although it is quite suitable as a quick-sort material. Biological sample (hair) results were above the minimum detectable activity (MDA) for only one of the tests. Several questions as to the best methods for sample handling and processing remain.