We describe the language features of dream narratives from 3 large samples of normal persons using Linguistic Inquiry Word Count (LIWC), a computer text analysis program. Compared with LIWC norms from waking narratives, LIWC dream narratives showed more use of function words in general, common words, past tense verbs, relativity (particularly space), inclusion, leisure, friend, and home words, and less use of second-person pronouns, present and future verbs, causation words, large words, and assent words. Dream narratives did not contain more negative emotion words. These patterns were consistent across investigators, samples gathered at different times from student and online sources, and instructions for dream reports (i.e., recent dream vs. important dream). Statistically significant correlations between dream language features and personality (as measured by the Ten-Item Personality Inventory and the Big Five Inventory) were few in number and small in effect sizes. We conclude with discussing the implications of computer text analysis of dreams in more systematic studies comparing linguistic features with dream themes in cross-cultural clinical populations, and the implications of these features for scientific understanding of the continuum of consciousness.