OBJECTIVES: The aims of the study are: first, to compare two short diagnostic instruments, CIDI-SF and CIS-R, with respect to the structured clinical interview for non-patient (SCID-I/NP) for anxiety and depressive disorders; and second, to evaluate the influence of four languages, Italian, Romanian, Spanish and French, on the concordance tests. METHODS: A total of 120 participants from Italy and Romania, 119 from Spain and 141 from France (N = 500) were recruited randomly in a local primary care research centre (GPs or medical centres). The instruments were administered during a unique session: the lay instruments by students in psychology and the SCID by experienced psychiatrists. Kappa, sensitivity/specificity/negative (NPV) and positive predictive values (PPV), ROC curve (AUC) and the Youden Index (Y) were calculated. RESULTS: Results were better for the CIDI-SF than the CIS-R for anxious disorders, depressive disorders and any of them. The results were identical to that obtained by the CIDI 3.0 for the three categories and constant across the languages except for depressive disorders in Romania and France. CONCLUSIONS: CIDI-SF is a cost-effective instrument and could be easily integrated into health surveys; its performance values are better across languages than values proposed by the CIS-R and could be increased by inclusion of few additional information.