OBJECTIVE: Employer Pitney Bowes lowered plan participant co-payments for oral diabetic medications, insulin and long-term asthma controllers on January 1, 2002 to promote participant adherence and potential medical cost savings. METHODS: A longitudinal retrospective analysis of pharmacy claims data was conducted to evaluate changes in plan participant adherence before and two years after the benefit change. Adherence was measured using the Adherence Index (AI), a published method based on medical-cost regression modeling of plan participant-level persistency and compliance with therapy as measured from pharmacy claims history. Full-time employees and dependents with diabetes and asthma were identified using medical and pharmacy claims algorithms. Pharmacy eligibility was required between January 1, 2001 and December 31, 2003. RESULTS: Potential users identified included 318 asthmatics (average age 28.6 years) and 297 diabetics (average age 48.28 years). Mean adherence significantly increased in the year after the benefit change for long-term asthma controller users (n=94) (baseline mean 0.76 (standard deviation (s.d.)=0.19); follow-up year one mean 0.82 (s.d.=0.14) paired samples t-test p<0.0001). Long-term asthma controller adherence in follow-up year two remained significantly higher than baseline but was not significantly different from follow-up year one. There was a significant shift in the percentage of asthmatic plan participants with suboptimal adherence (AI<0.80) from 20 percent to 5 percent (McNemar p value=0.003). A significant increase in follow-up year one adherence was also detected for insulin users in (n=46): (baseline mean 0.78 (standard deviation (s.d.)=0.15); follow-up year one mean 0.84 (s.d.=0.12) paired samples t-test p<0.0001). Adherence increases were not measured in the oral diabetic category which had higher baseline adherence (Mean=0.94, (s.d.=0.11). CONCLUSION: This analysis suggests there may be an association between a reduction in co-pay and plan participant adherence as measured in claims data.