Background. Most research on syndromes of chronic fatigue has been con
ducted in clinical settings and is therefore subject to selection bias
es. We report a population-based incidence study of chronic fatigue (C
F) and chronic fatigue syndrome (CFS). Methods. Questionnaires assessi
ng fatigue and emotional morbidity were sent to 695 adult men and wome
n who had replied to a postal questionnaire survey 1 year earlier. Pos
sible CFS cases, subjects with probable psychiatric disorder and norma
l controls were interviewed. Results. Baseline fatigue score, the leve
l of emotional morbidity and a physical attribution for fatigue were r
isk factors for developing CF. However, after adjusting for confoundin
g, premorbid fatigue score was the only significant predictor. A minor
ity of CF subjects, all female, had consulted their general practition
er; higher levels of both fatigue and emotional morbidity were associa
ted with consultation. Possible CFS cases reported similar rates of cu
rrent and past psychiatric disorder to psychiatric controls, but after
controlling for fatigue or a diagnosis of neurasthenia the current ra
tes were more similar to those of normal controls. Two new cases of CF
S were confirmed. Conclusions. Both fatigue and emotional morbidity ar
e integral components of chronic fatigue syndromes. The demographic an
d psychiatric associations of CFS in clinical studies are at least par
tly determined by selection biases. Given that triggering and perpetua
ting factors may differ in CFS, studies that examine the similarities
and differences between chronic fatigue syndromes and psychiatric diso
rder should consider both the stage of the illness and the research se
tting.