Snow and sea-ice albedo is known to be crucial for the heat exchange at high latitudes and altitudes, and additionally it is an important parameter in global circulation models (GCMs) because of its strong positive feedback properties. Even so, the way snow and sea-ice albedo is parameterised in GCMs today is strongly simplified. In this study, seven GCM snow albedo schemes, three GCM sea-ice albedo schemes, a thermodynamic sea-ice model and a completely databased multiple linear regression model are intercompared and validated against a large amount of validation data. The snow and sea-ice albedo cases are considered separately. Eight sites from Svalbard, the French Alps and six stations in the Former Soviet consisting of 59 years of point data are included for validating the snow albedo schemes. For the sea-ice counterpart, the scaling becomes more important, and three categories of data are used. The first category is point data from North Pole Drifting Ice Stations and SHEBA experiment, while the second is albedo data collected at a line in the SHEBA experiment. The last category consist of remote sensing validation data. For each of the available meteorological parameter, a 95% confidence interval is constructed, and the significant meteorological parameters for modeling the albedo are identified. Based on the significant regressors, the multiple linear regression model is constructed to include the parameters: temperature, snow depth, positive degree day, a dummy of snow depth and a constant for the snow albedo, and temperature, snow depth, a dummy of snow depth, cloud cover and a constant for the sea-ice albedo. It was found that the simulated snow albedo is more varying than the observed albedo for all snow models and -sites, and the modeled albedo is most often underestimated. The modeled snow albedo is decreasing at a faster rate or by a larger magnitude during winter snow metamorphosis than the observed data. Also, the decrease in snow albedo associated with snow melting is often delayed in the models for the Former Soviet sites. However, as we see it, the sea-ice albedo is too crude parameterised in the GCMs. This is easily seen when comparing the observed and modeled sea-ice albedo. To overcome this, we have suggested an improved sea-ice albedo parameterisation consisting of the four surface types and their corresponding -fractions: snow on sea-ice, bare sea-ice, melt pond and open water.