<p>Stroop effects can be influenced by context-specific cues associated with varying levels of proportion congruency, even for items that occur with equal frequency in each context. This theoretically important result points to the involvement of a cue-driven memory process in the dynamic updating of selective attention. However, attempts to replicate this result have been challenging, potentially due to limitations or confounds in the experimental design, or difficulties in inducing context-specific learning due to item-specific cues. The present study aimed to reevaluate the memory-guided hypothesis using the trial-unique Stroop paradigm, where each trial consists of a unique stimulus and response that never reappear throughout the experiment, thereby preventing trial-to-trial associative regularities. One context was associated with high conflict (20% proportion congruency), and the other with low conflict (80% proportion congruency). Experiment <InternalRef RefID="Sec2">1</InternalRef> demonstrated that trials within the low-conflict context exhibited larger congruency effects compared with those in the high-conflict context, aligning with the memory-guided selective attention hypothesis that attentional priorities become associated with contextual cues and are automatically reinstated through memory retrieval. Experiment&#xa0;<InternalRef RefID="Sec11">2</InternalRef> replicated this finding and examined whether both auditory and visual context cues were necessary by assigning participants to combined auditory–visual, auditory-only, or visual-only conditions. The context-specific proportion congruency effect was successfully replicated in both the combined and visual-only conditions, but not in the auditory-only condition, suggesting that visual contextual cues may be more effective than auditory cues. These findings provide compelling evidence for memory-guided selective attention while revealing important boundary conditions regarding contextual cue effectiveness.</p>