OBJECTIVES: To estimate the difference in responsiveness to drug benefit changes between sick and healthy populations under different drug benefit plans. METHODS: Prescription spending in 2001, demographic, co-pays and chronic condition information of 107,710 primary participants between ages 18 to 64 was obtained from Caremark PBM claims system. Existence of chronic conditions identified through pharmacy claims utilization algorithms was used as a proxy for sick people. A 2-part model was used to estimate the prescription spending on copay changes. The first part used a logistic regression to estimate the probability of incurring of any prescription. The second part used an OLS regression on log of total spending for the utilizing participants. RESULTS: Under a 1-Tier copay plan, if the co-pays increase from $5 to $10, the reduction of total spending is 6.4% more for sick people than for healthy people. Under a 2-Tier plan, if the co-pays increase from $5 to $10 for generic and $10 to $15 for brand, the reduction of total spending is 10.2% more for sick people than for healthy people. Under a 3-Tier plan, if the co-pays increase from $5 to $10 for generic, $10 to $15 for brand formulary and $15 to $20 for brand non-formulary, the reduction of total spending is 10.4% more for sick people than for healthy people. CONCLUSIONS: Based on the specific data set, this research shows that sick people are more responsive to drug benefit changes. Since cost-sharing designs are frequently used to contain total drug spending, understanding drug benefit designs for heterogeneous populations is crucial to achieve the optimal balance of sponsor savings and participants' health.