This study investigated feature- and dimension-based intertrial effects in visual search for a pop-out target. The 2 prominent theories explaining intertrial effects, priming of pop-out and dimension weighting, both assume that repeating the target from the previous trial facilitates attention shifts to the target, whereas changing the target leads to attentional switch costs. In contrast, the results from the present study indicate that intertrial effects from changing features and dimensions involve different underlying mechanisms: Eye movement recordings showed that feature priming reliably modulates the speed of visually selecting the target, whereas changing the target dimension interferes only with processes after selection of the search target. Further experiments with a size and orientation singleton target showed that feature priming does not consist in carryover effects of target activation or nontarget inhibition, contrary to standard assumptions in visual search. Instead, priming effects critically depended on whether a coarse relation between target and nontarget features (e.g., smaller or larger) was repeated or reversed across trials. These results suggest the need to modify current models of priming in visual search.