Sets of 16 beliefs about the consequences of each of 6 leisure activities were selected in a pilot study by means of free elicitation. The beliefs varied in salience (salient vs. nonsalient), valence (positive vs. negative), and type (instrumental vs. effective). In the main study, 69 college students expressed their agreements with each belief, evaluated the consequences involved in the beliefs, and rated each leisure activity on 4 evaluative scales as a standard measure of attitude. The latencies of these responses were used to assess accessibilities of beliefs and attitudes in an effort to quantify degree of salience. Accessibility of beliefs was found to increase significantly with frequency of elicitation in the pilot study (salience) and with the judged favorability of the consequences. Salience was further found to be related to degree and extremity of agreement. Accessibility was also found to moderate the correlation between beliefs and attitudes posited in the expectancy-value model of attitude. The correlation between a given belief and the direct attitude measure increased with the belief's accessibility, and the correlation between a measure of attitude based on the set of salient beliefs and the standard attitude measures increased with the accessibility of the attitude. Implications for the identification and measurement of perceived benefits and costs of leisure are discussed.