Abstract
The issues of low prevalence in the context of the diagnosis of neuropsychological disorders in the larger community is discussed. Guidelines are proposed for producing reliability and validity estimates which are clinically, as well as statistically meaningful.