An experiment to determine the most accurate and repeatable method for generating instrument inter-calibration functions (ICFs) is described, based upon data collected with a dual-beam GER1500 spectroradiometer system. The quality of reflectance data collected using a dual-beam spectroradiometer system is reliant upon accurate inter-calibration of the sensor pairs to take into account differences in their radiant sensitivity and spectral characteristics. A cos-conical field-based method for inter-calibrating dual-beam spectroradiometers was tested alongside laboratory inter-calibration procedures. The field-based method produced the most accurate results when a field-derived ICF collected close in time was used to correct the spectral scan. A regression model to predict the ICF at a range of wavelengths was tested, using inputs of solar zenith angle, cosine of solar zenith angle and broadband diffuse-to-global irradiance ratios. The linear multiple regression model described up to 78% of the variability in ICF; the remainder of the variability was most likely due to complexities of instrumental behaviour in response to warm-up time, ambient temperature and environmental conditions at the time of measurement. Collection of ICFs using a stable laboratory source was shown to provide unsatisfactory results due to differences between lamp outputs and the field-measured solar spectrum. Consequently, the most practical and accurate method of deriving inter-calibration functions is to use field-derived ICFs, collected close in time and space to the data requiring correction.