It is often desired to extract more information from a test score than is available in a single number. The almost universal response to such desires is to divide the overall test score into subcomponents/subscores (e.g., math and verbal scores, reading fluency and reading comprehension, etc.). We summarize the rules governing the legitimate use of subscores and report on the frequency, in modern practice, that it is done correctly. In short, dividing up a test into subscores reduces its reliability, and subsequently, its validity. Using the military’s ASVAB test as an example, we show that the overall score is the only good predictor of later performance and the nine subtests are not effective in differentiating types of skills and knowledge.