This review aims to clarify the use of the term 'dissociation' in theory, research and clinical practice. Current psychiatric definitions of dissociation are contrasted with recent conceptualizations that have converged on a dichotomy between two qualitatively different phenomena: 'detachment' and 'compartmentalization'. We review some evidence for this distinction within the domains of phenomenology, factor analysis of self-report scales and experimental research. Available evidence supports the distinction but more controlled evaluations are needed. We conclude with recommendations for future research and clinical practice, proposing that using this dichotomy can lead to clearer case formulation and an improved choice of treatment strategy. Examples are provided within Depersonalization Disorder, Conversion Disorder and Posttraumatic Stress Disorder (PTSD).