OBJECTIVES: Until 1995 sulfonylureas were the only oral diabetes medications available in the United States. Currently, there are expanded options in diabetes therapy with five classes of oral medications. Our objective is to test whether the newer therapies diffuse more quickly in higher income areas among an insured population. METHODS: All prescriptions for oral diabetes medication issued to patients in metropolitan Baltimore between 1999 and 2001 were selected from CareFirst Inc.'s claims database. Records were aggregated to the patient level with indicators for all classes of drugs used. Patients were divided into those receiving only sulfonylureas and those receiving at least one of the newer medications. Each record was linked to census data for median income by zip code. Linear and logistic regressions were performed on microdata and data aggregated by zip code. Our null hypothesis is that income is not associated with use of newer therapies. We interpret a statistically significant coefficient on income as evidence to reject the null. RESULTS: The mean proportion of the population receiving sulfonylureas only declined from 41% to 28% over the three years. Regression results show a statistically significant negative association between income and monotherapy. Each $10,000 of income is associated with a 1% increase in the share of the population receiving newer medications. Controls for age and gender strengthen this association. Between 1999 and 2001 the coefficient associating income and likelihood of monotherapy increased from .003 to .011 (per $10,000). CONCLUSIONS: Within an insured population, diabetics residing in higher income neighborhoods are more likely to receive one of the new drug therapies. The level of this disparity increased over the course of our study. A subsequent paper will distinguish between co-pay effects and physician practice effects.