It is well known that contamination of platinum‐based thermocouples due to use at high temperatures causes the local Seebeck coefficient of the wire to change from its ‘initial’ state. As this exposure and contamination is usually not uniform along the length of the thermocouple, the Seebeck coefficient also becomes a function of position along the thermocouple, leading an exposure‐dependent thermoelectric signature or inhomogeneity. At NML (CSIRO) we have, for many years, included explicit measurement of the thermocouple inhomogeneity in all calibrations of ‘used’ thermocouples, using a special uniform‐temperature ‘scan furnace’ having an entrance region with a steep gradient. The thermocouple inhomogeneity value is used in the calculation of the calibration uncertainty: indeed it is usually the dominant component, as the uncertainty in the reference standard SPRTs and fixed points is usually negligible by comparison. The inhomogeneity measurements at NML are usually done at 450 °C, low enough to limit defect annealing drift, but high enough to generate a reasonable thermoelectric voltage. The inhomogeneity measured at 450 °C, expressed as a percentage of the total EMF, has been assumed to be representative of the inhomogeneity at other temperatures. To date, this assumption has not been thoroughly validated. In this paper, the thermoelectric signatures of several thermocouples with inhomogeneities caused by years of ‘real world’ use are examined at temperatures 250, 450, 650, 850 and 1000 °C, and compared with our traditional 450 °C homogeneity assessments. The results allow us to assess the validity of the assumption that inhomogeneity may be expressed as a percentage of the total thermo‐voltage.