The item-specific proportion congruency (ISPC) effect is the finding of attenuated interference for mostly incongruent as compared to mostly congruent items. A debate in the Stroop literature concerns the mechanisms underlying this effect. Noting a confound between proportion congruency and contingency, Schmidt and Besner (2008) suggested that ISPC effects are entirely contingency based. We introduce a broader theoretical analysis that points to the contribution of both contingency and item-specific control mechanisms. Our analysis highlights that proportion congruency is not confounded with contingency when the relevant dimension functions as the ISPC signal, and predicts that evidence of item-specific control should be obtained by shifting the signal from the irrelevant to the relevant dimension. We examine this prediction in a picture-word Stroop paradigm. When the relevant dimension functions as the ISPC signal (Experiments 1 and 2), evidence of control is obtained. When the irrelevant dimension functions as the ISPC signal (Experiment 3), contingencies can account for the ISPC effect. These patterns support our theoretical analysis, challenge a pure contingency account, and favor the inclusion of control in accounts of ISPC effects.