The retreat of Arctic sea ice extent during summer has attracted considerable attention from the public and science communities, especially since the record shattering minimum set during the summer of 2007. The estimates of sea ice extent are primarily based on retrievals of sea ice concentration (SIC) from passive microwave satellites. These estimates of SIC suffer from contamination from the overlying atmosphere and differences in surface emissivity between first-year and multi-year sea ice. Aerial radiometer data and ship observations have been used to validate SIC estimates and find that these errors vary seasonally. Melt-ponding, water vapor, clouds, wind, rough sea surface, and instrumentation error all contribute to errors in the SIC analyses, and as a result, SIC retrievals from passive microwave tend to be lower than the in-situ observations. Weather systems cause large brightness temperature fluctuations over short time scales spanning a few days. To address these issues, we have been developing numerical weather filters to improve the analysis of SIC from Scanning Multichannel Microwave Radiometer, Special Scanning Microwave Imager, and Advanced Microwave Scanning Radiometer to produce more accurate fields of SIC. Preliminary results show that our methods are able to remove SIC analysis errors due to changes in clouds, relative humidity, and varying surface conditions. Briefly, the weather filter improves SIC estimates due to passing weather systems by: 1) removing spurious SIC estimates over areas of open water; 2) increasing SIC estimates under clouds, and 3) decreasing SIC estimates under relatively dry air. Over first-year sea ice, the corrections to the SIC estimates range from –10% to as much as 30% during summer, while during winter and over multi-year sea ice the corrections are on the order of ±10%. These numerical weather filters may be used to improve retrievals of SIC for the Antarctic, and retrievals of surface temperature.