Formal diagnostic systems (DSM-IV, ICSD, and ICD-10) do not provide adequate quantitative criteria to diagnose insomnia. This may not present a serious problem in clinical settings where extensive interviews determine the need for clinical management. However, lack of standard criteria introduce disruptive variability into the insomnia research domain. The present study reviewed two decades of psychology clinical trials for insomnia to determine common practice with regard to frequency, severity, and duration criteria for insomnia. Modal patterns established frequency (> or =3 nights a week) and duration (> or =6 months) standard criteria. We then applied four versions of severity criteria to a random sample and used sensitivity-specificity analyses to identify the most valid criterion. We found that severity of sleep onset latency or wake time after sleep onset of: (a) > or =31 min; (b) occurring > or =3 nights a week; (c) for > or =6 months are the most defensible quantitative criteria for insomnia.