This paper discusses methodological issues arising in the initial stages of a larger epidemiological case-control study. Practitioners from both Generic Mental Health and Substance Misuse Services (n = 170) were asked to identify which of their clients, from a time-limited caseload (n = 2341), had comorbid mental health and substance misuse problems. Although practitioners were provided with a definition of 'singly diagnosed' and 'dually diagnosed', it became apparent that these definitions were applied pragmatically, depending on the nature of the client's primary problem and the agency they were presenting to. Issues raised include the time period in which a client was considered to have a concurrent mental health problem and substance misuse, how a 'mental health problem' was defined and whether a personality disorder should be categorized as a 'mental health problem'. There was also some disagreement about whether clients who were being treated primarily by Substance Misuse Services, but were also taking prescribed antidepressants, implicitly had a 'mental health problem'. We raise these methodological issues, as they have implications for determining the prevalence of 'dual diagnosis' and the subsequent provision of services.