Feature singleton search is faster when the target-defining dimension repeats across consecutive trials than when it changes (Found & Muller, 1996). However, this dimension repetition benefit (DRB) has also been demonstrated for the tasks with no search component (Mortier, Theeuwes, & Starreveld, 2005). If DRBs in the search and non-search tasks have the same origin, significant DRBs across trials of different tasks should rise. Two different tasks varied either in a predictable manner (Experiment 1) or randomly (Experiment 2) across trials. In detection task, search displays containing either color or orientation singletons were used. Discrimination task required identification of either color or orientation of a single presented item (non-search display). In Experiment 3, participants performed only the discrimination task, while the search and non-search displays varied randomly. There were significant DRBs for both tasks when the task repeated but not when the task changed (Experiments 1 and 2). DRBs were significant both when the display type repeated and when it changed (Experiment 3). Overall, the findings can be well explained by assuming multiple, independent dimension-weighting systems generating DRBs in different tasks. (Contains 4 footnotes, 1 table, and 5 figures.)