Test indices are often determined by comparing test results of healthy persons with test results of patients known to have the disease. However, the patient population for which the test is ordered in clinical practice often differs from the study population on which the test indices are based. Hence, these indices are not applicable to clinical practice and should be recalculated using data from daily clinical practice. Two major problems of using routinely collected data are discussed: the assessment of the final health status and tracing the reason for ordering the test. Prior considerations are given to the use of hospital information systems (HIS) to sample the patient population that is desired and to collect the necessary data for calculating test indices. We investigated whether the HIS of Leiden University Hospital (which is presented as an example) can be used to calculate the indices of clinical laboratory tests, histopathologic examinations and radiodiagnostic investigations. The results indicate that the registration of diagnoses must be improved and that a way must be found to capture the implicit reasoning for ordering diagnostic tests.