This chapter evaluates different types and different approaches to simulation in computational neuroscience, including small and large-scale simulations, and bottom-up and top-down approaches. It outlines what kind of explanations simulations can provide, including a brief discussion of mechanistic vs computational explanations. The chapter also outlines some interesting benefits of computational simulations in this field to computer science, in the form of neuromorphic devices. Simulations are a key way to investigate brain function, and most simulations focus on small-scale 'local circuits' to investigate a specific cognitive function. These include simulations of sensory processing, decision making, motor control, and different types of learning and memory. Two different approaches to large-scale simulation that attempt to address some of these problems are reviewed below. One is a 'bottom-up' approach which attempts to address the problem of biological realism within simulations. The second is a 'top-down' approach which attempts to address the issue of whole-brain function across distributed local circuits.