Scope – The Collaborative Spectrum Project aims to define subthreshold and atyical conditions not sufficiently characterized in the current diagnostic nomenclature and for which adequate assessment instruments are not available. This paper reports on the development and validation of new instruments to assess the spectrum of five psychiatric disorders. Design – Three multicenter studies and one single-site study were conducted in Italy to assess the validity and reliability of the five spectrum interviews. Another cross-sectional study to validate the panic-agoraphobia spectrum has been conducted in Pittsburgh. Setting – Outpatients attending various university clinics, university students and, in one Italian study, gym attenders were recruited for the studies. Main outcome measures – Five structured clinical interview to assess the spectrum of panicagoraphobia (SCI-PAS), mood (SCI-MOODS), social phobia (SCI-SHY), and the obsessive-compulsive (SCI-OBS) and eating disorder spectra (SCI-ABS) were administered along with a diagnostic interview and a number of self-report and interviewerrated instruments. Results – All the domains of the interview showed high test-retest reliability (intraclass correlation coefficient >0.61) and satisfactory internal consistency. Mean domain scores were significantly higher in cases than in controls and in patients with the disorder of interest than in patients with other disorders. Convergent validity was satisfactory for panic-agoraphobia, social phobia and obsessive-compulsive spectrum domains. Differences emerged between SCI-ABS and self-report instruments assessing eating disorders. A cut-off score for the panic-agoraphobia spectrum was defined and its clinical validity was tested. Conclusions – The psychometric properties of the five spectrum interviews are very satisfactory, and studies are currently ongoing to test the clinical validity of all the spectra. Subthreshold and atypical symptoms deserve attention in epidemiological investigation.