This work discusses a snowfall detection approach for high-latitude regions that is based on a combination of passive sensors to discriminate between snowing and nonsnowing areas. Two different techniques have been developed to compute the probability of detecting a snowing event. The first technique is based on a logistic distribution to represent the probability of snowfall given the predictors; the second is a Bayesian technique not requiring any hypothesis as to the shape of the probabilistic model. The spaceborne Advanced Microwave Sounding Units A and B (on NOAA-16) and the Moderate Resolution Imaging Spectroradiometer (on Aqua) have been collocated and used for the estimation. Products from the spaceborne cloud-profiling radar on CloudSat are used as truth to calibrate and validate the proposed approaches. A comparison with a wellknown rain-rate model, developed by the Remote Sensing and Estimation Group of the Massachusetts Institute of Technology, is also presented, showing that both proposed methods discriminate snowing and nonsnowing condition over the polar regions, reducing by at least 30% the false alarms while considerably increasing the probability of detection.