OBJECTIVES: To determine the predictive validity of eight different adherence measures by studying the variability explained between each measure and two outcome measures: hospitalization episodes and total non-pharmacy cost. METHODS: This study was a retrospective analysis of the Arkansas Medicaid administrative claims data. Schizophrenia (ICD-9-CM = 295.**) patients were identified in the recruitment period July 2000 through April 2004. Patients had to be >18 years old, have prescription benefits, and continuously eligible for 6 months prior and 24 months after their first antipsychotic prescription. Persons taking two different antipsychotic drug products simultaneously were excluded. Medication Possession Ratio (MPR), Proportion Of days Covered (PDC), Medication Possession Ratio, modified (MPRm), Refill Compliance Rate (RCR), Compliance Ratio (CR), Continuous Measure of Medication Gaps (CMG), Days Between Fills Adherence Rate (DBR), and Continuous, Single Interval Measure of Medication Acquisition (CSA) were computed and modeled to predict hospitalizations and non-pharmacy costs. RESULTS: A total of 3,971 patients with a mean age 46.5 years, 54% were female, 55% were white, and 38% were black. The one-year adherence rates were numerically similar for MPR (0.74 SD= 0.31) and PDC (0.72, SD=0.29). Other measures such as MPRm (82.21%, SD= 27.27), RCR (97.20 % SD=115.05), CR (81.39% SD = 64.18) DBR (97.20% SD=115.05) and CSA (0.95 SD=0.79) had higher adherence values. The partial R-squares of the multivariate regression models will be reported. CONCLUSION: There is considerable variability in rates of adherence among the 8 measures of adherence. MPR and PDC which used the entire study period as the denominator had the most conservative adherence values. In contrast MPRm, PDC, RCR, CR, DBR which only consider the time between first and last prescription fill dates estimate higher levels of adherence.