Researchers too frequently consider the reliability of the scores they analyze, and this may lead to incorrect conclusions. Practice in this regard may be negatively influenced by telegraphic habits of speech implying that tests possess reliability and other measurement characteristics. Styles of speaking in journal articles, in textbooks, and in professional standards and guidelines are explored. Two recommendations are offered. First, the statement "the test is reliable" should be recognized as being inappropriate, and professional standards and editorial guidelines should make this clear. Second, an important implication of the realization that reliability inures to data, rather than tests, is that reliability should generally be explored whenever data are collected. Three tables present language usage examples. An appendix lists 52 articles surveyed. (Contains 28 references.) (Author/SLD)