An analysis of ten-minute albedo variations, recorded on Haut Glacier d'Arolla, Switzerland, over an 11 day period in the 1999 ablation season is presented. Most of the short-term (<1 day) albedo variability is caused by variations in cloud cover, while solar zenith angle variations in the range 25° to 75° are of minor importance, probably due to the predominantly cloudy conditions during the measurement period. A new method to calculate albedo variation as a function of cloud cover is developed. Short-term albedo variations are expressed by the ratio of the measured albedo to the daily albedo 'minimum', defined as the albedo under cloud-free conditions when the solar zenith angle is <50°. Variations in cloud cover are quantified by the ratio of the measured incoming shortwave radiation flux to the theoretical direct-beam shortwave radiation flux. The resulting relationships are successful, explaining 83% and 87-90% of short-term albedo variation on snow and ice respectively, and may be incorporated into albedo parameterizations already used in numerical energy balance melt models, without the need for additional data. Simulations with a glacier energy balance model suggest that melt rates are overestimated by between 1 and 3 mm water equivalent per day if a correction is not made for the increase in albedo under cloudy conditions. Other causes of albedo variation are identified and evidence is found for the removal of fine debris from the glacier surface by intense rainfall, leading to an albedo increase. The implications for energy balance models and satellite-derived albedo measurements are discussed.