Future methods of estimating and forecasting snow pack dynamics for large river basins are likely to incorporate mesoscale meteorology forecast models. This meteorological data type is spatially continuous over a landscape, and thus is compatible with distributed modeling approaches. Additionally, the meteorological models can provide all the parameters needed to drive an energy balance method of snow pack conditioning and melt. Before mesoscale meteorology models are used to drive operational snow melt models, it will be necessary to assess various scaling issues. In this preliminary assessment, we evaluate the sensitivity of timing and magnitude of melt water flux to a range of meteorological cell sizes. Data from the NCAR/PENN State MM5 model were used to generate the atmospheric boundary conditions for our snow model. SNTHERM, a complex one-dimensional energy balance model that takes into account most physical processes within the snow cover, was used for snow pack computations. The snow model computations were distributed over an 80-km segment of non-mountainous, but hilly terrain of the Sava River basin in Bosnia-Herzegovina based on slope and aspect. The original data provided by the Air Force Global Weather Center (AFGWC) was at 10 km resolution. We scaled the data to coarser resolutions of 20-km, 40-km and 80-km and compared melt water time series. Preliminary results indicate that for larger meteorological cell sizes melt water flux may occur over a shorter time period.