Wintertime precipitation has a large impact on the Northeastern United States each year. Within any given storm, there is potential for banded snow events, some of which can produce heavy amounts of snowfall. Previous studies have shown that heavy banded snow events require strong values of frontogenesis, weak moist symmetric stability, and moisture to form within any given storm. This study seeks to find answers to two questions. The first is whether it is possible to predict mesoscale snow banding events based on independent variables. The second question is whether these variables can be combined into a forecasting tool to predict the occurrence of snow banding through statistical correlation. For this study, storms were first identified using GOES satellite archive data, obtained from the NCDC GIBBS archive (http://www.ncdc.noaa.gov/gibbs). Storms seen on satellite imagery were gathered into a database. Using WSR-88D radar data from the NCDC, banding events were identified in the storm database with the storms containing changes in 24 hour snow depth of 8 inches or greater. Using classification systems developed by Novak et al.(2004), the type of banding was identified within storms. A gridded area was created over the Northeastern United States with 20 km spacing in between grid points. The grid was rotated 45 degrees to better align with the axis of the Appalachian Mountains. Topography for the region of study was classified based on elevation using the Integrated Data Viewer (IDV) data source of the National Geophysical Data Center (NGDC) ETOPO1 dataset. Heights were interpolated to the grid points developed in the 20 km resolution grid. Basic classification schemes involving elevation were employed to correlate the existence of a banding event with the topographical classification. Model data from the North American Regional Reanalysis (NARR-A) was obtained and used in General Meteorological Package (GEMPAK), with data interpolated to the 20 km resolution grid. Results from synoptic composites based on banding type observed in storm events will be presented, along with predictability of banding by variables such as topographic slope, frontogenesis, cross-shore potential temperature gradient, and the existence of a coastal front in the gridded domain. To statistically analyze the predictor set determined from the synoptic composites, a best subsets analysis was generated to determine the best variables to use in a regression equation. All three variables (topographic slope, frontogenesis, and cross-shore potential temperature gradient) were used in the regression equations for each banding type. Results of the best subsets analyses were used to determine the best variables to use in the linear regression analyses. These results are shown in Chapter 5. Linear regression analysis was also completed for all banding types following the best subsets analysis. The results of this test showed that the generated equation could not predict the occurrence of a banding event. In order to improve on this, other techniques should be used as means for analysis of banding. This indicates the need for inclusion of additional predictors, due to the influence of uncontrolled variables in the analysis. These results and future suggestions are discussed in Chapter 5.