Forecasts of snow depth are available as direct output from the Hirlam model. However, the treatment of frozen water in the physical parameterisation of the model is rudimentary: in particular, precipitation reaching the ground is regarded as snow if the temperature of the lowest model layer is below freezing. Clearly, this is an inadequate criterion. We may therefore ask whether forecast guidance on the occurence of snow can be obtained in some more reliable way. In this report we describe the application of some empirically derived criteria to the model output fields. Maps depicting the conditional probability of snow, in the event of precipitation, are prepared using these criteria. Their use is illustrated by consideration of a particular snowfall event in the Dublin area.