To explain the phenomenon that certain English verbs resist passivization (e.g., "*£5 was cost by the book"), Pinker (1989) proposed a semantic constraint on the passive in the adult grammar: The greater the extent to which a verb denotes an action where a patient is affected or acted upon, the greater the extent to which it is compatible with the passive. However, a number of comprehension and production priming studies have cast doubt upon this claim, finding no difference between highly affecting "agent-patient/theme-experiencer" passives (e.g., "Wendy was kicked/frightened by Bob") and non-actional "experiencer theme" passives (e.g., "Wendy was heard by Bob"). The present study provides evidence that a semantic constraint is psychologically real, and is readily observed when more fine-grained independent and dependent measures are used (i.e., participant ratings of verb semantics, graded grammaticality judgments, and reaction time in a forced-choice picture-matching comprehension task). We conclude that a semantic constraint on the passive must be incorporated into accounts of the adult grammar.