The purpose of the study was to determine such cut-off points in the scores of General Health Questionnaires (GHQ-12 and GHQ-28) that allow for optimal identification of people with mental health disorders in the Polish working population attending primary health care settings. The groups under the study covered 419 and 392 patients for GHQ-12, and GHQ-28, respectively. In the GHQ-12 group, 90 and in the GHQ-28 group, 80 subjects filled in the questionnaires and agreed to participate in the second stage of the study--a psychiatric interview. The criterion validity of the GHQs was a mental health diagnosis, based on the Munich version of Composite International Diagnostic Interview. The complete computerized version of interview, covering all diagnostic sections, has been adopted. In the mental health diagnosis only disorders, which currently troubled patients were taken into consideration and disorders which created problems in the distant past were excluded. In the group covered by GHQ-12 examination, 55.6% of persons had at least one type of mental disorder diagnosed, based on the criteria of both Diagnostic and Statistical Manual for Mental Disorders (DSM-IV) and the International Classification of Diseases (ICD-10). In the GHQ-28 group, the percentage of persons with mental disorders was 47.5%. After excluding patients with nicotine dependence disorder only, the frequency of mental health problems decreased to 45.5% and 33.8%, respectively. The proposed cut-off points, 2/3 points for GHQ-12 and 5/6 points for GHQ-28, were established at the level of the highest possible sensitivity and specificity not lower than 75%. These principals have been accepted for a practical reason, as the acceptance of the lower level of specificity forces medical practitioners to devote too much time to practically healthy people. At the above mentioned cut-off points for GHQ-12 sensitivity is 64% and specificity--79%, while for GHQ-28 the values are 59% and 75%, respectively. These validity coefficients were calculated from distributions of groups, from which persons with nicotine dependence as the only disorder were excluded. Incorporation of these people in the whole sample reduced the questionnaires' validity. Modification of responses scoring from the standard one--GHQ to CGHQ has not improved the validity of questionnaires. Lower validity coefficients of GHQ-28, in comparison to GHQ-12 validity are the effect of greater influence of somatic disease on the results acquired in this scale version of the questionnaire.