Eysenck (1984, Bulletin of the Psychonomic Society, 22, 545-548) suggested that storage characteristics may be an important determinant of worry, and postulated that prolonged worry occurs in individuals who have tightly organised clusters of worry-related information stored in long-term memory. These clusters reflect areas or domains of worry. Because the information is stored in tight clusters, it becomes more accessible, more rapidly activated and therefore retrieved more quickly. The Worry Domains Questionnaire (WDQ) (Tallis, 1991c) is used to determine which domain worried subjects most (Primary) and least (Secondary). Two experiments are reported using a word allocation task, which requires subjects to make categorical decisions, based on these worry domains. It is reported that priming facilitates the emergence of domain effects, thus providing support for a structural hypothesis. High worries take longer to reject negative words if they are from the Primary domain and have difficulty rejecting Primary domain words when they are under a congruent heading. In addition, high worriers are reported to show retarded latencies when attempting to process ambiguous information, consistent with Metzger et al.'s studies (1990, Journal of Clinical Psychology, 48, 76-88). It is suggested that the initiation and maintenance of worry is largely attributable to an elevated evidence requirement and this may link to the personality trait of perfectionism.