OBJECTIVES: Research has shown that Medicaid enrollees in the USA are the most frequent users of the Emergency Department (ED) services. Several studies have demonstrated that a high proportion of the ED visits could be avoided. The purpose of this analysis is to examine  the demographic  and health system delivery characteristics that are associated with avoidable ED visits (AEDV) among a high risk, high cost Medicaid population between 2008 and 2009. METHODS: One year claims dataset of a sample of high cost, high risk Medicaid enrollees in Houston, Texas was used for the analysis. This was design following the Andersen-Aday theoretical framework for studying access to health care. ED visits were classified into avoidable or not using the New York University algorithm. Patient complexity was measured using the Chronic Illness Intensity Index (CI3), an index used  to measure need of case management intensity. We performed logistic regression models to test for significant association between AEDV, and population at risk and health care delivery characteristics. RESULTS: We found that 69% (179) of our population had an ED visits during 2008-2009. Of these visits, 60% were classified as AEDV. The analysis showed that women were 33% less likely to have an AEDV per month. Age was negatively associated, with younger patients being more likely to have AEDV. More complex patients were 6.6% more likely to have an AEDV. For every extra physician a patient visited, the probability of having an AEDV per month increased by 2.4%, however this was not significant (p =0.06) at 95% confidence interval. CONCLUSIONS: Among high cost, high risk Medicaid patients there are certain patient characteristics that can allow us to identify those at higher risk of having an AEDV.  This information could be use to identify groups that would benefit from interventions to reduce ED utilization.