Previous research has suggested that, in visual-search tasks, the comparison between target and display items does not require attentional capacity. In the present experiment we used a secondary-task paradigm to distinguish the amount and duration of the attentional demands of visual search. The subjects performed visual search (the primary task) and tone detection (the secondary task) concurrently over the course oftive experimental sessions (1,440 trials). For each subject, target-response mapping was either consistent or varied for Days 1-5. The results indicate that the amount of attentional demand, as reflected in secondary-task performance, increased as a function of display size in the search task. Switching from consistent to varied mapping in a sixth experimental session increased both the amount and the duration of the attentional demands of the search. The present results support models of visual-search performance in which the comparison of target and display items requires attentional capacity.One of the most prominent characteristics of visualsearch perfonnance is the influence of the number of items in the display on reaction time (RT) and error rate. Typically, visual search becomes slower and less accurate as the number of display items increases. As Logan (1976, 1978) has pointed out, however, this display-size effect is an empirical fmding that is compatible with alternative theories of visual-search performance. The display-size effect may represent the involvement of limited attentional capacity in the comparison between target and display items (Atkinson, Holmgren, & Juola, 1969;Rumelhart, 1970). Alternatively, the processing of display items may not require attentional capacity, but may involve a decision component with noise level being a function of display size (Estes, 1972;Gardner, 1973).Although the display-size effect alone is not sufficient to distinguish these different interpretations of visualsearch performance, relevant evidence can be obtained from dual-task versions of visual search, in which search and some other task are performed concurrently. Dualtask methodology assumes that concurrent tasks compete for one or more pools oflimited attentional capacity (Le., processing resources); attentional demands are consequently revealed in the comparison between dual-task and single-task levels ofperfonnance (Wickens, 1984). One example of dual-task methodology is the secondary-task paradigm, in which subjects are instructed to devote most of their attention to one of the tasks (the primary task) and to perform the other (secondary) task with whatever "reserve" attention is available. Under ideal conditions (i.e., the two tasks do not require the same sensory-motor systems and primary-task performance is not disrupted by the addition of the secondary task), changes in secondary-task perfonnance reflect the attentional capacity demands of the primary task (Duncan, 1980; Kantowitz, 1985;Kerr, 1973). Logan (1978) proposed that the secondary-task paradigm can provide information regar...