OBJECTIVES: In order to evaluate the costs of different health states for patients with cardiometabolic risk, a study was performed to operationalize these health states, to identify individuals from a claims database and assign them to the health states. METHODS: Claims data of a German sickness fund with 1.5 million beneficiaries were used for the years 2000 to 2004. Only patients aged 18-80 years were included who were continuously covered by the health insurance during this period. Health states were composed of different attributes (diabetes mellitus with/without micro- or macrovascular complication, hypertension, hypercholesteremia, hypertriglyceridemia, coronary heart disease, obesity), each of them being transposed into appropriate ATC codes or ICD-10 codes (inpatient, outpatient, sick leave diagnosis). Patients were selected from the database according to their health state pattern. RESULTS: Out of n=774,132 beneficiaries (62% male), n=736,653 (95%) could be assigned to one of the defined health states. Most of them (58%) were allocated to the health state without any of the defined attributes. 27% had 1 to 4 cardiometabolic risk factors, but no diabetes. Four percent had no diabetes, but had already experienced cardiovascular diseases such as myocardial infarction and/or stroke. 6% matched one of the diabetes related health states. CONCLUSION: Typical limitations of any analysis performed on the basis of claims data should be borne in mind. These comprise lack of diagnostic accuracy and incomplete knowledge of the patients' case histories and clinical measures for severity of illness. Nonetheless, claims data provide useful information for economic modelling, as they derive from a naturalistic setting and allow an unbiased view on health care delivery and utilization under real-life conditions. This can determine the authenticity of economic models.