The validity and reliability of an empirically defined fatigue syndrome were tested in a prospective cohort study of 245 primary care patients, with glandular fever or an upper respiratory tract infection. Subjects were interviewed three times in the 6 months after onset. Subjects with the empirically defined fatigue syndrome were compared with those who were well and those who had a psychiatric disorder. The validity of the fatigue syndrome was supported, separate from psychiatric disorders in general and depressive disorders in particular. Only 16% of subjects with the principal component derived fatigue factor also met criteria for a psychiatric disorder (excluding pre-morbid phobias). Compared with subjects with psychiatric disorders, subjects with the operationally defined fatigue syndrome reported more severe physical fatigue, especially after exertion, were just as socially incapacitated, had fewer mental state abnormalities, and showed little overlap on independent questionnaires. A more mild fatigue state also existed. Both fatigue syndrome and state were more reliable diagnoses over time than depressive disorders. The empirically defined syndrome probably is a valid and reliable condition in the six months following glandular fever.