Background. A cardinal feature of schizophrenia is the sufferer's difficulty in interacting appropriately within the social milieu. This deficit has recently been associated with the concept of theory of mind, more commonly construed as a working model to understand behavioural patterns in autistic children. In this paper the complex relationships between theory of mind, IQ and psychoses are addressed.Methods. Five experimental groups were used; non-psychiatric controls, affective disorder, schizophrenia with normal pre-morbid IQ, schizophrenia with pre-morbid IQ in the mildly learning disabled range, and mild learning disability with no history of psychiatric illness. All subjects were given a first order Theory of Mind Task and if successful, a second order Theory of Mind Task was then administered. All subjects were rated using the Positive and Negative Symptom Scale (PANSS).Results. Subjects with schizophrenia and subjects with mild learning disability show impaired ability on a second order theory of mind task. However, when patients who are unable to answer reality questions are removed from the analysis specific impairment of theory of mind is only seen in subjects with schizophrenia. Furthermore, this impairment is relatively specific to particular psychopathological clusters in subjects with schizophrenia. Even though the same clusters of psychopathology are also seen in patients with affective disorder, their presence is not associated with poor second order theory of mind performance.Conclusions. Impaired theory of mind on second order tests is specific to schizophrenia when compared to mild learning disability and affective disorder control groups. Subjects with schizophrenia and pre-morbid mild learning disability show greater impairment than subjects with schizophrenia and a pre-morbid IQ within the normal range.