Contemporary cognitive models incorporate an information-processing approach in explaining the causes of depression. The common core in these models is that depression is caused by a vicious cycle of negative thoughts and maladaptive behaviours, and maintained by an excessive focus on the causes and consequences of the depression symptoms (i.e. rumination). Using a variety of study designs, this thesis examines three aspects of cognition assumed to underlie this process: negative attentional bias, cognitive control and metacognitions. The thesis shows that using a computerized intervention to target negative attentional bias in currently remitted depression patients leads to reduced interaction between depression and anxiety symptoms, and that this is associated with increased interest and motivation for social involvement. Furthermore, the thesis indicate that there are bidirectional links between depression symptoms, rumination, and negative beliefs about rumination (negative metacognitions) in currently remitted depression patients, suggesting that negative metacognitions have a relevant role in the feedback loop between depression and rumination. The thesis also demonstrates that there is a basic association between negative metacognitions and reduced ability to shift between mental sets, suggesting a fundamental interaction between cognitive control processes and metacognitions. In sum, the findings presented in this thesis are in line with the notion of depression as a complex interplay of several cognitive and emotional mechanisms within and between different levels of processing.