The extent to which archaeological or cemetery skeletal collections accurately represent the population from which they were drawn cannot be known. The creation of documented or forensic skeletal collections, derived from donation or autopsy, was intended to overcome many of the problems inherent in archaeological populations, yet it is misleading to assume such collections represent a specific or defined population. This study compares the documented skeletal collection curated at the Maxwell Museum to annual demographic information from three relevant populations: (i) the living population of New Mexico (NM), (ii) the deceased of NM, and (iii) the subset of decedents who undergo a medicolegal death investigation or autopsy. Results indicate that the Maxwell Documented collection differs significantly from all three populations in every variable examined: age, sex, ethnicity/race, cause, and manner of death. Collection development that relies on body donation or retention of unclaimed bodies under coroner/medical examiner statutes results in a biased sample, with significant overrepresentation of males, Whites, the elderly, those who die unnatural deaths and individuals with antemortem traumatic injury or surgical intervention. Equally problematic is the perception that the collection has documented race or ethnicity, when in fact only 17% was self-reported, while the affinity of the remaining individuals was determined by pathologists or other observers. Caution is warranted in how this and similar collections are used and interpreted by researchers. Although documented reference collections are useful in developing methods of estimating age or sex, they are not a proxy for modern or racially/ethnically defined populations. Am J Phys Anthropol, 2008. © 2008 Wiley-Liss, Inc.