Conservation genetics has expanded its purview such that molecular techniques are now used routinely to prioritize populations for listing and protection and infer their historical relationships in addition to addressing more traditional questions of heterozygosity and inbreeding depression. Failure to specify whether molecular data are being used for diagnosis-related questions or for population viability questions, however, can lead either to misinterpretation of character data as adaptive information or to misinterpretation of frequency or distance data as diagnostic or historical information. Each of these misinterpretations will confound conservation programs. The character-based approach to delimiting phylogenetic species is both operationally and logically superior to 'diagnostic' methods that involve distance- or frequency-based routines, which are unstable over time. Tree-based criteria for the diagnosis of conservation 'units' are also inappropriate because they can depend on patterns inferred without reference to diagnostic characters. Intraspecific studies, conservation-related or otherwise, that adopt terminology and methods designed to infer nested hierarchic relationships confuse diagnosis with historical inferences by treating diagnoses as outcomes rather than as precursors to phylogeny reconstruction. A character-based diagnostic approach recognizes the analytical dichotomy between species hierarchies and population statistics and provides a framework for the understanding of each. No species concept, however, should be viewed as an absolute criterion for protecting populations, but as part of a framework from within which identification of protection and management goals can be achieved effectively and defensibly.