OBJECTIVES: Prescription drugs that provide high value should command higher prices than lower value drugs other things equal. We examined correlations between a drug's price and its economic merit, as measured in cost-utility analysis (CUAs). We also examined whether formularies policies are consistent with cost-utility (CU) ratios. METHODS: CUAs from 1998 through 2001 on pharmacotherapies were selected from a large registry of analyses. All CU ratios and drug cost estimates were taken as reported in analyses and standardized to 2002 US$. Spearman correlation coefficients were used to quantify the association between drug prices and CU ratios. We examined the Florida Medicaid Preferred Drug List and Harvard Pilgrim Pharmacy Program to analyze whether insurers cover drugs with good value. Wilcoxon rank sum test was performed to assess if preferred drugs had different ratios than non-preferred drugs. RESULTS: Of 205 ratios, 16.1% were for short-term treatment (<2 months), 29.8% intermediate treatment (2-18 months), and 54.2% lifetime treatment. Ten and seven-tenths percent of ratios were cost saving and 8.3% dominated. Correlations between ratios and prices were 0.4991 (p = 0.0069), 0.1154 (p = 0.4724) and 0.2892 (p = 0.0041) for short, intermediate, and lifelong drugs, respectively. CU ratios did not differ significantly from preferred to non-preferred drugs on both health plans. Among cost-saving therapies, only 68.2% were covered by Florida Medicaid and 72.7% by Harvard Pilgrim. Among dominated drug interventions, 88.2% were covered by Florida Medicaid and 94.12% by Harvard Pilgrim. CONCLUSIONS: CU ratios of pharmacotherapies are positively associated with price, but the correlation is low among intermediate and lifelong treatments. Preferred drugs on two health plans generally do not reflect better cost-effectiveness. These results may reflect the absence of value-based pricing and lack of evidence-based reimbursement policies, or the fact that CUAs are poorly conducted (e.g., they have inappropriate comparators), or do not reflect decision makers' perceptions of value.