Cost of illness estimates for chronic diseases can be underestimated if only costs related to diagnosis and treatment of that disease are measured. This underestimation results from overlooking costs associated with secondary consequences of the disease such as complications and co-morbid conditions associated with the disease. OBJECTIVE: Two methods for estimating medical care costs of diabetes were compared: an "attributable" method" and a "case-control" method. METHODS: The study population was all diabetic patients in the 1999 Medical Expenditure Panel Survey, a nationally representative series of probability surveys on the use and cost of medical care in the United States. "Attributable" costs were estimated by summing costs specifically associated with diabetes. "Case-control" costs were estimated by subtracting costs between diabetic cases and non-diabetic controls which were matched on age, gender, race, and number of comorbid conditions not related to diabetes. Costs were summarized for pharmacy, hospital inpatient, outpatient, and emergency room care and reported in 1999 dollars. RESULTS: The total cost of illness was $3046 per patient using case-control method compared to $1151 per patient using the attributable method. The case-control method found costs to be higher for all categories of care, with the largest being hospital inpatient costs. Cost differences were statistically significant for all categories except for emergency room care. CONCLUSIONS: Diabetic "attributed" costs accounted for only 39% of the total difference in health care costs between diabetics and matched controls. Patients with diabetes use more medical services than controls, but a large portion of this care is not specifically attributed to diabetes.