The main aspects of the most common models describing depression in chronic pain patients are reviewed. It is suggested that dualistic thinking provides neither a satisfactory model of chronic pain, nor of depression, and relies on questionable assumptions of homogeneous, diagnostically defined entities. Models of depression based in cognitive psychology, although apparently more suitable, cannot be applied to populations of pain patients without clarifying the relationship between pain and depression. Furthermore, commonly used depression measurement instruments are criticized for criterion contamination, lack of external reference, and lack of sensitivity when applied to these groups, all of which further obscure the relationship. Finally, we suggest more promising directions for research in this area.