The perceptual load theory resolved the debate between early and late selection theories, such that the perceptual load of a given task determines attentional selection, with the selection occurring earlier when the load is high and later when the load is low. Flanker congruency effect caused by processing of a task-irrelevant distractor has been used as the indicator for choosing between early or late selections, such that the flanker effect was significant under low load condition but was none under high load condition. However, according to the conflict monitoring theory, the state of conflict control in the current trial is influenced by the conflict condition of the previous trial. It is then of theoretic importance to investigate whether the sequential effect of conflict resolution is affected by the perceptual load of the task at hand. In the present study, participants were asked to search for a target orientation (a vertical or a horizontal bar) and to make discriminative response among homogeneously (low load condition) or heterogeneously (high load condition) oriented non-target bars in the central display, which was flanked by a congruent or an incongruent bar presented at the left or the right periphery. All trials were pseudo-randomized to evaluate the sequential impact of the perceptual load and the congruence of the previous trial (trial n-1) on the attentional selection and conflict resolution in trial n. Results showed that the flanker interference effect, in terms of the reaction times (RTs) in the congruent condition subtracted from the incongruent condition, was significant for the current trial low perceptual load condition, but was none for the current trial high load condition. Moreover, the perceptual load and the flanker congruence in trial n-1 affected the RTs and the flanker interference effect in trial n. When trial n-1 was of high perceptual load, trial n showed typical perceptual load effect, i.e. larger flanker effect for low load condition but none flanker effect for high load condition. However, when trial n-1 was of low load and incongruent condition, flanker effect in low load condition of trial n was significantly reduced, with no difference with the flanker effect in high load condition. When trial n-1 was of low load and congruent condition, the flanker effect in high load condition of trial n remained non-significant, but the effect in low load condition of trial n was significantly enhanced. These results indicated that the early or late attentional selection in the current trial is not completely determined by the current perceptual load, but is also affected by the state of attentional selection and conflict resolution of the previous trial. Reduced cognitive control in the previous trial causes more interference effects from task-irrelevant information, especially when there are spare attentional resources left from processing the current task.