OBJECTIVES: Antipsychotic non-adherence has been associated with poor health outcomes in patients with chronic comorbid conditions.  We explored the relationship between antipsychotic adherence and adherence with medications for chronic comorbid illness in the North Carolina Medicaid program. METHODS: Medicaid claims were used to identify continuously enrolled patients with schizophrenia from 2001-2003.  The annual prevalence of comorbid diabetes, hyperlipidemia, and hypertension was evaluated among these patients.  The relationship of antipsychotic adherence with comorbid medication adherence was examined descriptively using the proportion of days covered (PDC).  In addition, patients with comorbid illness who experienced a 90 day gap in antipsychotic treatment were compared to patients without a 90 day gap.   Difference-in-difference regression was used to compare comorbid medication adherence 10 months pre- and post-antipsychotic gap between patients with and without a 90 day antipsychotic treatment gap.  RESULTS: Of the 6841 continuously enrolled Medicaid patients identified with schizophrenia in 2001, the prevalence of comorbid hypertension, hyperlipidemia, and diabetes was 13.9%, 16.6%, and 18.9% respectively in 2002 and 15.3%, 20.0%, and 21.0% respectively in 2003.  Descriptive trends suggested a correlation between antipsychotic and chronic comorbid medication adherence.  Patients with a 90 day antipsychotic gap experienced a decline in adherence to comorbid medication following the gap.  In comparison to patients without a gap, patients who discontinued antipsychotic treatment for 90 days or more experienced a 4.5% [95% Confidence Interval: (-16.0, 7.1)], 1.5% (-6.7, 3.65), and 4.3% (-8.6, -0.04) reduction in antihypertensive, antihyperlipidemic, and antidiabetic PDC respectively following the gap. CONCLUSIONS: While non-adherence is a behavioral phenomenon affecting both the treatment of schizophrenia and comorbid conditions, we observed an incrementally larger decline in comorbid medication adherence among patients with antipsychotic treatment gaps.  Larger sample sizes, additional control for confounding, and assessment of corresponding health outcome measures are needed to further explore the magnitude of this problem.