The authors review the constraints of current mental disorder classification systems that rely upon descriptive symptom-based approaches, and weigh the benefits and hazards of these classification and diagnostic strategies. By focusing principally on superficial descriptions of symptoms, current systems fail to address the complex nature of persons' transactions within and adaptations to difficult environments. While attempting to be atheoretical, current systems exclude types of information that may elucidate individuals' functioning across various contexts, often because it is difficult to obtain such data reliably. With current approaches, misdiagnosis is likely, particularly when diagnostic criteria are applied to persons in nonclinical settings. Alternative approaches that take fuller advantage of clinicians' expertise and other forms of clinical data are reviewed, and recommendations are made for the next generation of classification systems. Application of evolutionary theory to psychiatry and psychology, as well as development of a theory and nosology of context in terms of persons' adaptations, are needed to expand our knowledge of normal and abnormal human development and psychopathology.