Two studies comparing patients with Parkinson’s disease (PD) and controls on flanker tasks are discussed. Study 1 investigated effects of PD on instantaneous (within-trial flanker congruency) and sequence-dependent (between-trial congruency sequence) distractor interference. For situations requiring instantaneous cognitive control, PD patients and controls showed similar congruency effects on reaction time (RT) and on event-related brain potentials (ERPs). As expected, controls showed reliable congruency sequence effects on RTs and ERPs. However, PD patients were completely unaffected by the congruence sequence across consecutive trials, as revealed by RTs and ERPs. Study 2 examined persistent (input selection) versus transient (input shifting) mechanisms of attention control in PD using a novel combination of a flanker task with an attentional set-shifting task. Controls showed robust shifting costs (prolonged RTs), but PD patients did not show evidence for comparable shifting costs. PD patients made more errors than controls when required to shift between attentional sets, but also when they had to initially maintain an attentional set. At the neural level it was found that contrary to controls, PD patients did not display any ERP augmentation on shift trials. Our results reveal that disturbed contextual modulation of cognitive control is a major cognitive dysfunction in PD patients.