Injection of large volumes of carbon dioxide (CO2) for the purposes of greenhouse-gas emissions reduction has the potential to induce earthquakes. Operators of proposed projects must therefore take steps to reduce the risks posed by this induced seismicity. In this paper, we examine the causes of injection-induced seismicity (IIS), and how it should be monitored and modelled, and thereby mitigated. Many IIS case studies are found where fluids are injected into layers that are in close proximity to crystalline basement rocks. We investigate this issue further by comparing injection and seismicity in two areas where oilfield wastewater is injected in significant volumes: Oklahoma, where fluids are injected into a basal layer, and Saskatchewan, where fluids are injected into a much shallower layer. We suggest that the different induced seismicity responses in these two areas are at least in part due to these different injection depths. We go on to outline two different approaches for modelling IIS: a statistics based approach and a physical, numerical modelling based approach. Both modelling types have advantages and disadvantages, but share a need to be calibrated with good quality seismic monitoring data if they are to be used with any degree of reliability. We therefore encourage the use of seismic monitoring networks at all future carbon capture and storage (CCS) sites.