Depression is a debilitating and recurrent mental health problem. Although there are a number of effective psychological treatments for adult depression, around 50% of individuals do not recover (Cuijpers et al., 2021). To improve the effectiveness of these treatments we need to understand how they work. Previous research has identified times in treatment when there are patterns of discontinuous depression change and these times have been used to examine processes of change to further understand how treatments lead to depression change. The aim of this thesis was to build upon this research to further understand discontinuous depression changes in and outside of treatment, the processes of change surrounding these times of depression variability, and how they relate to treatment outcomes. This thesis primarily focused on two patterns of discontinuous change; rapid improvements in depression symptoms, known as ‘sudden gains’ (Tang & DeRubeis, 1999) and ‘depression spikes’ which are transient increases in depression symptoms (Hayes, Feldman, Beevers, et al., 2007). To examine this four studies were conducted. Study one investigated the rates, timing, and association with treatment outcomes of sudden gains and depression spikes in a large scale clinical practice dataset. Study two explored client cognitive and behavioural processes of change surrounding sudden gains in cognitive behavioural therapy (CBT) and behavioural activation (BA), and their association with treatment outcomes in a trial dataset. Study three used the same trial dataset to explore predictors of depression spikes in CBT and BA, and their relation to treatment outcomes. Study four focused on how cognitive and behavioural avoidance are associated with depression variability outside of treatment across a stressful life period in a student sample. The thesis ends with a discussion of the methodological, theoretical, and clinical implications of the findings and suggestions for future research.