In previous analyses of data from the present general population study we found that screening of anxiety and depression symptoms by the Hopkins Symptom Checklist-25 (HSCL-25) and diagnostic classification by the Composite International Diagnostic Interview (CIDI) identified the same amount of cases, but agreed in only half of them. In this paper we compared and validated the screening cases with the classificatory cases by the use of medication, loss of functioning and help seeking (illness indicators). We thought that the CIDI cases would have more illness indicators, because they reflected diagnoses, "true illness", in contrast to the HSCL-25, which was a more unspecific measure of distress. The HSCL-25 and the illness indicators data were collected in a stage I random individual population sample above 18 years during 1989-1991 (N = 1879, response rate 74%), the CIDI data were collected in a selected stage II, (N = 606, response rate 77%). The stage II data were weighted to represent the population sample. Screening cases by the HSCL-25 had significantly more illness indicators than diagnostic cases by the CIDI. Cases agreed upon with both instruments had the most illness indicators, cases agreed upon only by the CIDI had the least. Diagnoses give information about help eventually needed, the HSCL-25 distress measure expresses more the urgency with which it is needed. The choice between the HSCL-25 and the CIDI would depend on the aim and the resources of the study. If evaluation of needs is involved, using an instrument picking up both classification and distress would be the best choice. Given our positive experience with interviewing with the CIDI, a CIDI improved to be more sensitive to how much distress a certain diagnosis exerts on the individual would be a good choice.