Traditional two-channel microwave radiometers (MWRs) are widely used to measure cloud liquid water path (LWP); however, the retrieved LWPs are subject to relatively large uncertainties, particularly for low LWP clouds. By reformulating the statistical retrieval method with clear-sky measurements as a reference, a simple method is presented to significantly reduce uncertainties in the LWP retrieval due to errors in MWR calibration, uncertainties in the absorption coefficients of atmospheric gases, and variations in the vertical profiles of temperature and pressure. The improvement is illustrated by comparing the statistics of the erroneous clear-sky LWP for the Department of Energy's Atmospheric Radiation Measurement Program Climate Research Facility observations at the North Slope of Alaska site and by comparing LWP retrieved with a multiple-sensor algorithm and LWP retrieved based mainly on MWR measurements. This letter also demonstrates the importance of using correct water cloud temperature and temperature-dependent water absorption coefficients for MWR LWP retrieval over cold regions. This approach can be easily implemented for combined MWR, ceilometer, and surface meteorological measurements.