OBJECTIVES: To develop predictive models for medication compliance in dyslipidemia that will aid healthcare decision makers in targeting compliance intervention programs. METHODS: Pharmacy and medical claims data from a commercial health plan were analyzed for all currently enrolled members who received their first dyslipidemic medication between May 1, 2007 and April 30, 2008. Percentage of days covered (PDC) defined as days supply of dyslipidemic medication per 365 days. PDC < 80% was used to categorize non-compliant patients. Potential independent variables included patient demographics, pharmacy utilization and medical conditions. Stepwise logistic regression was used to predict the odds of non-compliance. RESULTS: A total of 88,635 patients were included. Sixty-five percent of patients were non-compliant (PDC = 0.33; SD=0.22). The most significant predictor of non-compliance was treatment with bile acid sequestrants (OR: 6.75; p <0.0001, compared to statins). Significant predictors of non-compliance also included age category, increasing from an OR=1.11 for age 45-55 to OR= 3.23 for age < 18 [p <0.0001 for all estimates compared to age 75+]; prior diabetes diagnosis (OR: 1.15, p <0.0001) and the number of unique pharmacies used (OR=1.10 per additional pharmacy; p <0.0001). Factors reducing non-compliance include male gender (OR: 0.77, p <0.0001); previous heart attack (OR: 0.82; p = 0.0221); prior compliant behavior (OR: 0.888; p <0.0001); number of unique physicians seen for medications (OR: 0.969 per additional physician; p <0.0001) and copayment categories (relative to no copayment). Compliance significantly improved by 12%, 12% and 6% for copay categories $5-$10, $10-$20, and $20-$30, respectively to no copayment. (p<0.01). CONCLUSIONS: The results may be used by health care decision makers to identify patients who are most likely to be non-compliant with dyslipidemia therapy and help focus medication compliance efforts. A secondary analysis using corresponding lab data is projected.