This experiment investigated the properties affecting the time taken to read individual words and to discriminate between words and non-words. The number of letters in a word was found to affect the time taken to read it aloud, but not the time taken to discriminate it from non-words. The frequency of a word affected the time for discrimination, but not that for reading. Non-words which obeyed the rules of English orthography and phonology were more difficult to discriminate from words than those which violated those rules. Imageability, concreteness, and the number of syllables in a word were found not to affect performance, nor were derived nouns more difficult to process than simple nouns. It is suggested that reading aloud employs grapheme-phoneme translation based upon a letter-by-letter analysis of the stimulus; that discriminating words from non-words obeying the rules of English orthography and phonology employs a search of the lexicon based upon a holistic analysis of the stimulus; and that discriminating words from non-words violating those rules employs a direct test of the regularity of the stimulus based upon the combinatory rules of English orthography.