During the past century, the scope of mental health intervention in North America has gradually expanded from an initial focus on hospitalized patients with psychoses to outpatients with neurotic disorders, including the so-called worried well. The Diagnostic and Statistical Manual of Mental Disorders (DSM), Fifth Edition, is further embracing the concept of a mental illness spectrum, such that increasing attention to the softer end of the continuum can be expected in the future. This anticipated shift rekindles important questions about how mental illness is defined, how to distinguish between mental disorders and normal reactions, whether psychiatry is guilty of prevalence inflation, and when somatic therapies should be used to treat problems of living. Such debates are aptly illustrated by the example of complicated bereavement, which is best characterized as a form of adjustment disorder. Achieving an overarching definition of mental illness is challenging, owing to the many different contexts in which DSM diagnoses are used. Careful analyses of such contextual utility must inform future decisions about what ends up in DSM, as well as how mental illness is defined by public health policy and society at large. A viable vision for the future of psychiatry should include a spectrum model of mental health (as opposed to exclusively mental illness) that incorporates graded, evidence-based interventions delivered by a range of providers at each point along its continuum.