The performance of two meteorological pre-processors, which calculate the surface fluxes of heat and momentum from routine meteorological observations, was studied during two winter months at high latitudes, characterized by low solar altitudes and snow-covered surfaces, i.e. high albedo values. This, together with low temperatures and low wind speeds, causes stable stratification (often with inversion) to be predominant. The preprocessors studied are those in use by the Swedish and Finnish Meteorological Institutes. Comparisons were made between the pre-processors and between each one of the pre-processors and direct turbulence measurements of the fluxes. In addition, the net radiation predicted by the two schemes was compared with measurements. The results show that the calculated momentum fluxes agree quite well with the measurements, but that the pre-processors need to be further developed to give reasonable estimates of the surface heat flux in the studied conditions. The friction velocities calculated by the two pre-processors agree quite well with each other, but there is a large discrepancy between the two schemes for sensible heat flux. One reason for the difference between measured and calculated heat flux values is the parameterization of net radiation in the preprocessors. The net radiation estimated by the pre-processors was shown to differ from the measured values. This is not surprising, bearing in mind that the pre-processors have been calibrated against measurements at mid-latitudes and with grass-covered surfaces.