A key aspect of higher cognitive control is the ability to switch efficiently between alternative tasks in a manner that is appropriate to the situation. For instance, a bilingual speaker needs to control which language she/he uses depending upon who is being spoken to and to switch between languages when necessary. Since in any given context there is a range of tasks that could be performed, the question arises of how we control our use of different tasks over time. Much of the research into dynamic cognitive control for the past two decades has employed the ”task switching” procedure which requires participants to respond to stimuli according to one of two alternative task-rules and switch the appropriate task-rule from trial to trial. Researchers have consistently found in such procedure that participants require more time in switching between two tasks than in merely repeating one task. In this article, I will review studies that have been conducted for the past two decades with various forms of task-switching paradigms and by use of behavioral and neuroimaging approaches. The review focuses on preparatory control mechanisms (switch-specific and/or generic) and theoretical accounts in task switching. The review also focuses on various interference phenomena in task switching, including switch response-repetition cost, proactive interference of tasks, backward inhibition, task rule congruency effect, and competitor rule suppression. Finally, I will provide perspectives on future research for the next decade.