Background. Numerous nosological decisions are made when moving from the common human symptom of unusual fatigue to the rare chronic fatigue syndrome (CFS). These decisions have infrequently been subjected to rigorous evaluation.

Method. We obtained telephone interview data on fatiguing symptoms from 31406 individuals twins in the Swedish Twin Registry aged 42–64 years; 5330 subjects who endorsed fatigue and possessed no exclusionary condition formed the analytic group. We evaluated the definition and classification of CFS-like illness using graphical methods, regression models, and latent class analysis.

Results. Our results raise fundamental questions about the 1994 Centers for Disease Control criteria as (1) there was no empirical support for the requirement of four of eight cardinal CFS symptoms; (2) these eight symptoms were not equivalent in their capacity to predict fatigue; and (3) no combination of symptoms was markedly more heritable. Critically, latent class analysis identified a syndrome strongly resembling CFS-like illness.

Conclusions. Our data are consistent with the ‘existence’ of CFS-like illness although the dominant nosological approach captures population-level variation poorly. We suggest that studying a more parsimonious case definition – impairing chronic fatigue not due to a known cause – would represent a way forward.