People learn to associate external (predictive) cues (e.g., pictures; colors) with the attentional demands (e.g., the likelihood of conflict) that tend to accompany these cues. Such learning supports item-specific control, the reactive triggering of control settings associated with predictive cues (e.g., high level of focus triggered by a cue predicting high attentional demands). Item-specific control is assumed to operate with a degree of automaticity that allows for efficient processing even in the presence of competing demands. In three experiments, we investigated whether the unpredictable appearance of another salient stimulus (external distractor) presented along with the predictive cue would interfere with the triggering of item-specific control settings. The first two blocks of each experiment (i.e., acquisition phase) allowed participants to learn associations between different pictures and their likelihood of conflict in a picture-word Stroop task without external distraction. In the last two blocks (i.e., test phase), we introduced a random visual distractor (Experiments 1 and 2) or a combined visual and auditory distractor (i.e., multisensory; Experiment 3), with Experiment 2 additionally manipulating the timing of the distractor onset. Overall, the item-specific proportion congruence effect remained intact in both distractor-present and distractor-absent trials in all experiments, suggesting that item-specific control is robust to the presence of external distraction. We consider the theoretical implications of the results, with a focus on the automaticity of item-specific control and future investigations of potential boundary conditions.