There is an urgent need to identify objective biomarkers for the assessment of bipolar disorder, to improve diagnosis and prognostic evaluation. Neuroimaging is a particularly promising approach. We review here the structural and functional neuroimaging studies carried out on bipolar disorder. These studies have led to the development of neurobiological models of bipolar disorder assuming cortical-limbic dysregulation. Dorsal brain structures are thought to decrease in volume and activity in bipolar disorder, reducing inhibition of the ventral-limbic network and enhancing emotional responses. These models also assume abnormal prefrontal-subcortical limbic connectivity. This abnormal connectivity has been identified by both diffusion tensor imaging studies (anatomical connectivity) and functional MRI (functional connectivity). However, studies are currently limited by the heterogeneity of the patients included. Future research should include studies to validate biomarkers for the assessment of bipolar disorder and studies of large and well characterized samples of patients with bipolar disorder.