Chronic fatigue syndrome (CFS) is a heterogeneous illness characterized by a high prevalence of psychiatric problems. We reasoned that we could reduce heterogeneity by excluding patients with psychiatric problems preceding CFS. We compared the functional status, mood, fatigue level, and psychiatric status of this more homogeneous group of CFS patients with the same parameters in patients with mild multiple sclerosis and in patients with major depression or dysthymia. Patients with CFS and those with multiple sclerosis were similar in terms of level of anger, severity of depression, level of anxiety, and frequency of current psychiatric diagnoses. Patients with CFS resembled depressed patients in having impaired vigor and experiencing substantial fatigue and confusion--problems constituting part of the case definition of CFS. The group with CFS was not psychologically vulnerable before the development of this condition and maintained adequate networks of social support despite disabling illness. Stratification to exclude patients with prior psychiatric disease and those with mild CFS allowed us to define a group of patients with CFS who more resembled patients with mild MS than patients with major depression or dysthymia and thus were more likely to have illness with an infectious or immunologic cause. Use of such a stratification strategy should prove important in testing of the viral/immunologic hypothesis of the etiology of CFS.