The result of a glomerular filtration rate (GFR) measurement on a particular patient is of limited use to the referring physician since normal GFR values vary widely with the patient's age and build, etc. To overcome this problem, it is usual to normalise the measured GFR by dividing it by the patient's surface area and multiplying the result by the surface area of a 'standard' man. This transforms the measurement onto a scale which applies to all patients, young and old, large and small, where normal values fall within a well-defined range and where the degree of renal impairment can be quantified. We have examined the generally accepted surface area (SA) and the less well-known extracellular volume (ECV) normalisation methods of GFR measurements in a series of 110 patients. The results show that both methods produce essentially the same result; however, ECV normalisation is theoretically more correct, can be found directly without the patient's ECV being measured and does not require the use of empirical formulae. Mathematical justification for ECV normalisation is presented, and a proposed distribution pattern for the normalised measurement is introduced. A simple mathematical model shows that accurate GFR measurements can be made in the presence of an enlarged ECV, but normalisation of these will produce misleading low values.