Functional magnetic resonance imaging (fMRI) is a powerful tool that can be used to examine the role of brain regions and networks in the pathophysiology of psychiatric disorders. This dissertation applies a range of methodological techniques to fMRI data to better understand mood and anxiety disorders. In Chapter 1, we demonstrated the malleability of reward-related brain function, which has been shown to be aberrant in major depressive disorder (MDD) and bipolar spectrum disorders (BSD). More specifically, we showed that in healthy adults, positive mood enhances activity in reward-related brain regions specifically during reward anticipation. In Chapter 2, we examined brain activity and connectivity related to anhedonia, a core symptom of numerous psychiatric disorders. We showed that trait anhedonia in healthy adults is negatively associated with activation in brain regions involved in processing reward and salient emotional stimuli. We also used psychophysiological interaction (PPI) analysis to show that trait anhedonia is associated with reduced connectivity of mesolimbic, and related limbic and paralimbic networks involved in reward processing. In Chapter 3, we demonstrated that anhedonia in MDD is associated with context-specific deficits in posterior ventromedial prefrontal cortex connectivity with the reward network when encountering pleasurable stimuli, rather than a static deficit in intrinsic resting-state connectivity. Analyses for this study included generalized PPI and seed-based resting-state connectivity. In Chapter 4, we developed a novel method to assess how connectivity changes as a function of stress and arousal in healthy adults. Here, we showed that the salience network is able to optimally engage the executive control network to coordinate cognitive activity at moderate levels of arousal, but is unable to do so at high levels of arousal and stress. Finally, in Chapter 5 we examined BSD, high-risk, and healthy control adults, and used machine learning to predict reward/loss versus intrinsic brain states based on features of large-scale brain networks. In addition to achieving classification accuracies above 83%, we showed that the discriminability of reward/loss versus intrinsic brain states was related to self-report measures of reward sensitivity. Overall, this dissertation used a range of methodological approaches to better understand the neural underpinnings of mood and anxiety disorders.