"Anxious depression" is used variably both by researchers and clinicians to describe admixtures of anxiety and depressive symptoms. The authors sought to determine the best model for conceptualizing anxious depression by studying a sample of depressed patients referred to a tertiary referral unit. Anxiety and depression were assessed using a comprehensive set of mixed symptoms that were subsequently refined to provide separate anxiety and depressive factors, and patients were trichotomized into groups of low, medium, and high anxiety on the basis of their total anxiety factor scores. Associations between the constructs of anxiety and depression in different depressive subgroups were explored, and the severity of depressive symptoms and other clinical variables across the three anxiety groupings was assessed. Depression variables were not linearly associated in a consistent pattern with anxiety-defined groups, arguing against a simple interdependence model driven by a higher-order variable such as depression severity. By contrast, the state anxiety categories were linked strongly with lifetime anxiety disorder prevalence, with some associations linear and with others evidencing a trend break association. The authors found support for a model of anxious depression, whereby anxiety both predisposes to nonmelancholic depression and contributes to its presentation by shaping its clinical features. Such a model and its definition assist in clarifying the cause of anxious depression and its treatment.