The reference instruments to measure the surface UV irradiance are based on a double monochromator system. The spectral irradiance yielded by these instruments is influenced by temporal instabilities and nonlinearities in the signal, the cosine error as well as uncertainties introduced in the needed prior calibrations. In this paper, we have carried out an uncertainty analysis of the spectral irradiances measured by using a mobile spectroradiometer that complies with the requirements of the Network for the Detection of Atmospheric Composition Change (NDACC). The spectral measurements were performed on 9th June 2005 ( cloudless sky conditions) at the Izana Observatory (28.3. N, 16.5. E, 2367 m above sea level, Tenerife, Spain), during an international intercomparison campaign organized in the framework of the project Quality Assurance of Spectral Ultraviolet Measurements in Europe (QASUME). At solar zenith angles smaller than 30 degrees, we found that despite the variations due to wavelength shifts induced by the high environmental temperature the relative expanded uncertainty was about 7% in the UV-A part of the spectrum; an increment was observed at wavelengths shorter than 315 nm such that the expanded uncertainty of the UV-B irradiance at 300 nm wavelength was about 9%. The measurements of the other five teams that participated in the campaign were within the bound specified by these expanded uncertainties. We also found that, regardless of the influence of the cosine error, the uncertainties involved in the absolute calibration procedure accounted for about 65% of the irradiance uncertainty. Although only a double monochromator was used in this work, the methodology applied to evaluate the uncertainty is general and it agrees with recommendations of the ISO Guide to the Expression of Uncertainty in Measurement.