At the lower end of IQ distributions in general populations, there is a clear excess of cases, representing the distinct pathology of severe learning disability. This study aimed to establish whether such a subpopulation exists in distributions of common mental disorder and depression symptom scores, above epidemiological 'case' cut-offs. Data from 9556 non-psychotic respondents to the 1993 OPCS (Office of Population Censuses and Surveys) National Household Psychiatric Morbidity Survey were analysed. The distribution of total neurotic symptom and depression scores from the revised Clinical Interview Schedule were examined. Automated least squares methods were used to fit the best single statistical distribution to the data. A single exponential curve provided the best fit for the whole population, but floor effects produced deviations at symptom counts of 0-3. After truncation, exponential distributions fitted excellently. Proportions of the population above conventional cut-offs of > or = 12 symptoms differed by < 12% from expected for a range of low and high prevalence groups. The single exponential model also fitted the depression score. Symptom counts for the common mental disorders fall within single population distributions, with little apparent numerical excess in the case range. High and low prevalences of these disorders appear to be population characteristics, with shifts in exponential means predicting proportions above case cut-offs.