People can learn to ignore salient distractors that occur frequently at particular locations. This distractor-location probability-cueing effect has been attributed to learnt suppression of the likely distractor locations at a pre-selective stage of attentional-priority computations. An alternative, post-selective account would be that distractors are as likely to capture attention at frequent as at rare locations, but attention can be disengaged faster from distractors at frequent locations. Eye-movement studies confirm that learnt suppression, evidenced by a reduced rate of oculomotor capture by distractors at frequent locations, is a major factor, whereas the evidence is mixed with regard to a role of rapid disengagement (pro: Wang et al., 2019; contra: Di Caro et al., 2019). However, methodological choices in these studies limited conclusions as to the contribution of a post-capture effect. Using an adjusted design, here we positively establish the rapid-disengagement effect and demonstrate further processes contributing to probability cueing beyond the oculomotor capture and disengagement dynamics. Moreover, we examine statistical-learning effects not only for distractors defined in a different visual dimension to the search target (comparable with previous work), but also for distractors defined within the same dimension, which are known to cause particularly strong attentional capture. Of theoretical importance, we corroborate our previous finding of responses being slowed (on distractor-absent trials) to targets at frequent distractor locations only in the same-, but not the different-, dimension condition. Consistent with a pre-selective origin, this target-location effect already impacted the latency of the very first saccade.