We have developed a $k$-coefficient retrieval approach for Atmospheric Infrared Sounder (AIRS) observations, using AIRS cloud-cleared radiances (ACCRs) as input. This new approach takes advantage of the available ACCR, reduces computational expense, offers an efficient and accurate cirrus cloud retrieval alternative for hyperspectral infrared (IR) observations, and is potentially applicable to the compilation of a long-term cirrus cloud climatology from hyperspectral IR observations. The retrieval combines a lookup-table method coupled to a residual minimization scheme using observed cloudy and cloud-cleared AIRS radiances as input. Six AIRS channels between 766 and 832 $\hbox{cm}^{-1}$ with minimal water vapor absorption/emission have been selected, and their spectral radiances have been demonstrated to be sensitive to both cirrus cloud optical depth $(\tau_{c})$ and ice crystal effective particle size $(D_{e})$. The capability of the $k$-coefficient approach is demonstrated by comparison with a more accurate retrieval program, which combines the delta-four stream (D4S) approximation with the currently operational Stand-alone AIRS Radiative Transfer Algorithm (SARTA). The distribution patterns and the range of retrieved cloud parameters from the $k$-coefficient approach are nearly identical to those from SARTA$+$D4S retrievals, with minor differences traced to uncertainties in parameterized cloudy radiances in the $k$ -coefficient approach and in the ACCR. The $k$ -coefficient approach has also been applied to four AIRS granules over North Central China, Mongolia, and Siberia containing a significant presence of cirrus clouds, and its results are quantitatively compared to simultaneous Moderate Resolution Imaging Spectroradiometer/Aqua cirrus cloud retrievals. Finally, AIRS retrieved $\tau_{c}$ and $D_{e}$ are consistent with the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) and CloudSat derived values for semitransparent cirrus clou...