Epidemiologic studies of chronic fatigue syndrome (CFS) have been hampered by the absence of a specific diagnostic test, but with increasing interest in this disorder there has been a greater understanding of the risk factors, illness patterns, and other aspects of this multisystem disorder. Working case definitions have been developed for research purposes but they have continued to change over time and have not always been utilized precisely by various investigators. This has been a major factor in the widely varying estimates of prevalence rates, but two different studies using the same working definition and including a medical work-up have estimated the prevalence to be approximately 200/100,000. Clusters of CFS cases, which appear to be related to earlier reports of "epidemic neuromyasthenia", have attracted considerable attention and appear to be well documented, although investigated with varying methodology and often with dissimilar case definitions. Risk factors for cases occurring in clusters and sporadically appear to be similar, the most consistent ones being female gender and the co-existence of some form of stress, either physical or psychological. The prognosis of CFS is difficult to predict, although cases occurring as part of clusters appear to have a better prognosis as a group than sporadic cases, and those with an acute onset have a better prognosis than those with gradual onset. It is highly unlikely that there is a single agent, infectious or noninfectious, that is responsible for more than a small proportion of CFS cases and, at the present time, the risk factors for developing CFS appear to lie more prominently in the host rather than the environment.