Most college-student readers have difficulty in detecting the letter F in instances of the word OF embedded in a single statement. Throughout a series of five experiments designed to clarify the basis of these detection failures, their unique and robust nature was demonstrated. The detection failures persisted in spite of repeated attempts to detect the letters by subjects who, in separate conditions and experiments, first memorized or copied the statement, or who, for purposes of comparison, also detected the O in OF, the N in ON, or the F in IF, or who read the statement in a number of physical formats, which included lower and upper letter cases, scrambled syntax, unsegmented letter strings, and vertical (list) presentation. Although many of these manipulations significantly improved performance, none produced perfect performance or performance comparable to the detection of F in IF. Several hypotheses, including those of redundancy, unitization, and phonetic recoding, were tested as explanations of the detection failures: The hypothesis that received the strongest support was that of phonetic recoding. This hypothesis focuses upon the atypical pronunciation of F as/v/(as in the word OF), rather than as the more typical/f/. In short, this reading illusion was concluded to be, in large part, a result of the subjects’ scanning their acoustic rather than their visual images of the printed word.