Over the past ten years or so, discussion of vegetation change in rangeland science has emphasised event- driven or episodic processes, occurring on timescales measured in decades or longer. Management recommendations arising from this literature have stressed that management must also be event-driven. This paper cautions against the uncritical acceptance of such a world view into management philosophies. We conclude that for management purposes, appropriate models of change in rangeland systems should include a balance between the effects of infrequent, unpredictable events and the effects of more continuous processes, measured in timescales of years or less. This may involve explicit recognition of multiple timescales in a hierarchical model system. We arrive at these conclusions from a number of perspectives. Firstly, a substantial proportion of total demographic change in shrub populations occurs between events. Secondly, managers are best able to devise appropriate management strategies by a process of adaptive management. This can only be successful if the adaptive cycles have a short return time. Thirdly, it is important that managers think of change as being continuous. Mental models held by managers must acknowledge the value of continuous change. This provides the best opportunity for acquiring knowledge through experience and helps prevent management inertia when faced with an event outside previous experience. Finally, management can take best advantage of a given event by 'conditioning' the resource. This can be thought of as managing within states, say by building up a seedbank, and provides the opportunity to alter the probability of a given event occurring.