AbstractIndirect interactions are almost inevitable in any multi-species community. Understanding the implications of such interactions is a challenging task, in light of the very large number of ways species can be tied together in complex food webs. One approach to this complexity is to focus on strong interactions among a relatively small number (e.g. 3-6) of species interacting in defined configurations: community modules. In recent years, the discipline of community ecology has developed a substantial body of theory focused on such modules. Modules often clearly describe the basic features of empirical systems, particularly in simplified anthropogenic landscapes, and also help to isolate and characterize key processes driving the dynamics of more complex communities. In this chapter, we draw out a number of insights from ecological studies of modules which we believe are relevant to biological control. We emphasize in particular the module of 'shared predation', where a natural enemy attacks two or more species of prey. Theoretical studies suggest a number of 'rules of thumb', including: (i) the greatest risk to non-targets may occur from control agents that are only moderately effective on the target; (ii) targets with a high reproductive capacity can indirectly endanger non-targets; (iii) there can be transient phases of extinction risk for non-targets during the establishment phase of control agents, particularly for species with high attack rates; (iv) at a landscape scale, mobile agents can endanger the fate of non-targets at sites other than the area of control; (v) using specialist natural enemies can pose risks to non-targets, if there are generalist resident predators/parasitoids which can exploit these introduced agents. The theoretical models help to highlight circumstances when these effects should be particularly strong.