A robot’s object level can have many reasoning processes, so there might be many options for controlling object-level reasoning processes with metareasoning. Thus, we are faced with the following question: Which reasoning processes should the meta-level control? We might make this decision based on information about which reasoning processes are consuming the most resources or which ones are most likely to fail. This chapter discusses the options that are available for metareasoning approaches and policies, using examples from our own work and the literature to illustrate the techniques, and the advantages and drawbacks of different options. By making these decisions about the metareasoning approach, we are beginning to specify how the robot will perform metareasoning. Naturally, we might find that these choices do not work as expected, and we’ll have to iterate and revisit these decisions. This chapter describes many ways that a robot can use metareasoning to improve performance, reliability, and safety.