OBJECTIVE: To estimate direct and indirect costs of diabetes in Texas in 1992. RESEARCH DESIGN AND METHODS: For most direct medical costs, we relied on third party and provider billing databases, including Medicare, Medicaid, VA facilities, public hospitals, and others. The researchers identified people with diabetes in the respective databases, located all records of their care, and sorted records as clearly, probably, or probably not attributable to diabetes on the basis of principal diagnoses. In most cases, costs were valued as allowable or paid charges. Some medical costs, such as private insurance, were estimated from national data and state surveys. Indirect costs included current short- and long-term disability costs and the discounted present value of future costs of mortality. Disability estimates relied on National Health Interview Survey (NHIS) data and U.S. Department of Labor wage data applied to Texas. Mortality estimates were based on death certificates. RESULTS: Total costs clearly or probably attributable to diabetes among Texans in 1992 were estimated at $4.0 billion. Direct medical costs were ∼ $1.6 billion. Indirect costs were estimated at $2.4 billion. the largest direct costs were paid by Medicare. Most indirect costs were from long-term disability. CONCLUSIONS: This study demonstrates methods for conducting cost of illness studies at the state level. In a state like Texas, with a large and growing Mexican-American population, estimation of current and future economic costs of diabetes is vital for development of strategies to minimize social and economic consequences of diabetes.