Satellite passive microwave (PM) sensors have observed sea ice in Polar Regions and provided sea ice concentration (SIC) data since the 1970s. SIC has been used as a primary data source for climate change prediction and ship navigation. However, the accuracy of PM SIC is typically low and biased in summer. To provide more accurate information for climatic research and ship navigation, it is necessary to evaluate quantitatively the accuracy of PM SIC and to account for its errors. In this research, we evaluated the SIC data derived from PM measurements using four representative sea ice algorithms: NASA Team (NT), Bootstrap (BT), Ocean and Sea Ice Satellite Application Facility (OSISAF) hybrid, and Arctic Radiation and Turbulence Interaction STudy (ARTIST) Sea Ice (ASI). Analyses were performed for the Chukchi Sea in summer using KOrean Multi-Purpose SATellite-5 (KOMPSAT-5) Enhanced Wide-swath synthetic aperture radar (SAR) images. Ice/water maps were generated by binary classification of texture features in the SAR images based on Random Forest, a rule-based machine learning approach. SIC values estimated from the sea ice algorithms showed good correlation with those calculated from the KOMPSAT-5 ice/water maps, but the root mean square error was larger than 10%. SIC values estimated from the algorithms showed different error trends according to the KOMPSAT-5 SIC range. All algorithms overestimated SIC values in open drift ice zones (KOMPSAT-5 SICs ranged from 0% to 15%). In marginal ice zones (SICs ranged from 15% to 80%), the OSISAF SIC values were the least biased compared to those from KOMPSAT-5. The NT algorithm largely underestimated SIC values in marginal ice zones, while the BT and ASI algorithms overestimated them considerably. All algorithms, except for BT, underestimated SIC in consolidated pack ice zones (SICs ranged from 80% to 100%). By analyzing the correlations of biases of SIC from the algorithms with the numerical weather prediction (NWP) data from the European Reanalysis Agency Interim reanalysis, it was found that the overestimation of NT and ASI SICs was largely influenced by atmospheric water vapor content, while the underestimation of NT and OSISAF SICs was owing to ice surface melting. The overestimation of BT SICs was not significantly correlated with the NWP data. The underestimated SIC from the BT and ASI algorithms for high SIC regions might be compensated by the atmospheric water vapor content. The differences in SIC values estimated from each algorithm were due to different sensitivities to atmospheric water vapor content in the regions with KOMPSAT-5 SIC lower than 40% and to ice surface melting in the regions with higher KOMPSAT-5 SIC.