Dreams are often defined as sleeping experiences with phenomenal character similar to perceptions of the real world. Hence they pose a prima facie challenge to accounts of phenomenal character in terms of acquaintance relations. One response is disjunctivist: to give a different account of their phenomenal character from that of successful perceivings. I argue that, given the alleged frequency of dreaming on the standard model, this disjunctivist approach weakens the explanatory value of the acquaintance account of the phenomenal character of successful perceivings. Another response is to follow Malcolm and Dennett in denying that dreaming has phenomenal character at all. I present a cultural-social model of dreams and argue that we lack theory-neutral evidence of the phenomenal character of dreams and thus it is legitimate to choose between theories of dreaming on the basis of their fit with our best theory of the phenomenal character of successful perceivings, namely acquaintance.