Setting and achieving realistic restoration goals requires careful consideration of what the goals should be, how achievable they are in the face of changing environmental conditions, and how best to reach the desired goal. Goals are often stated in either compositional or functional terms, or as a mixture of both.  Often it is assumed that getting the composition of a restored system right will also result in the restoration of desired functions. However, restoration of function does not necessarily entail returning to a pre-existing composition. Under global change scenarios it is no longer appropriate to assume that attempts to return a system to a pre-existing composition will either be possible or result in desired functional characteristics. How then should restoration goals be framed? The long running debate over the ecosystem function of biodiversity resulted in the realization that the types of species involved, rather than the number of species per se, were important in determining function. Hence in restoration it seems appropriate to consider using functional traits as a guide to deciding which types of species need to be replaced in a restoration effort. Deciding on which suite of functional traits are important in particular situations may then provide a set of options for restoration. Rather than adhering to a historically-authentic set of species under all conditions, it may be useful to consider an alternative array of species which encompass a desired set of functional traits: for instance including an array of disturbance responses or drought tolerance traits to confer resilience in the restored system. Such an approach may provide a pragmatic bridge between those wishing to remain as true to the historical authenticity of a site as possible and those taking a more functional approach in the face of changed environmental conditions.