Factor analysis of psychotic symptoms frequently results in positive, negative, and disorganized dimensions, but heritability estimates have not yet been reported. Symptom dimensions are usually only measured in individuals with psychotic disorders. Here, it is valuable to assess influences acting via liability to psychosis and independent modifying effects. We estimated heritability for psychotic symptom dimensions, taking account of these issues. Two-hundred-and-twenty-four probandwise twin pairs (106 monozygotic, 118 same-sex dizygotic), where probands had psychoses, were ascertained from the Maudsley Twin Register in London (1948-1993). Lifetime history of DSM-III-R psychotic disorder and psychotic symptom dimensions was assessed from clinical records and research interviews and rated using the Operational Criteria Checklist. Estimates of heritability and environmental components of variance in liability were made with structural equation modeling using a causal-contingent common pathway model adapted for ascertainment from a clinical register. Significant heritability was found for DSM-III-R psychotic disorder (h² = 90%, 95%CI 68-94%) and the disorganized symptom dimension (h² = 84%, 95%CI 18-93%). The heritability for the disorganized dimension remained significant when influences acting through liability to psychosis were set to zero, suggesting that some influences on disorganization are modifying factors independent of psychosis liability. However, the relative extent of modifying factors versus influences acting through psychosis liability could not be clearly determined. To our knowledge, this study provides the first formal evidence of substantive heritability for the disorganized symptom dimension, and suggests that genetic loci influencing disorganization in individuals with psychoses are in some cases different from loci that influence risk of psychotic disorders themselves.