Most definitions of severe mental illness (SMI) are categorical and assign the patient to either SMI or not-SMI status. While this is useful for some purposes, it is a rather limited approach. The purpose of the present study is to develop a new method of addressing the issue of 'severity', and to develop a dimensional rather than a categorical approach. The paper reports on the acceptability, reliability and validity of a method developed to collect a standard set of data covering the majority of items specified in the academic and policy literature as characterising SMI.