Traditional models of the relationship between cognition and emotion have typically presented the relationship between cognition and emotion as a single level of sequential processes. However, a number of more recent models have argued to the contrary that the relationship is complex and has to be modelled by multi-level processing systems. One such model, the SPAARS approach (Power & Dalgleish, 1997), is summarized, in particular, in relation to clinical theory and practice in the cognitive behaviour therapies. For example, the proposal in SPAARS that there are two parallel routes to the production of emotion has a number of interesting clinical consequences. Highlights are presented of what some of these consequences might be, and a number of recommendations are made for clinical practice.