Cloud altitude is an important parameter in determining the radiative impact of clouds on climate. In addition, atmospheric temperature and moister retrievals of cloudy scenes using high spectral resolution data require the cloud altitude be known. This paper will present a new algorithm to retrieve cloud top pressure using high spectral resolution infrared measurements that does not require clear sky simulations. The algorithm uses the CO2 absorption band between 650-800 wavenumbers. A selected clear sky scene is sorted relative to channel brightness temperature and a sorted index is created. The result is a smoothly increasing function of brightness temperature starting with the coldest most opaque channels to the warmest transparent channels in the CO2 band. In essence, the sorting has ordered the channels by the atmospheric level at which the channels weighting function is peaked. When the sorted index is applied to a cloud filled field of view the cloudy sorted spectrum will deviate from the clear sky sorted spectrum at channels whose weighting functions peak at or bellow the cloud attitude. The brightness temperature of this inflection point is related to the cloud emitting temperature. Using an atmospheric temperature profile the cloud top attitude can be inferred. Because this technique does not require a fast clear sky model, the implementation and computational expense is reduced compared to the CO2 slicing technique. The CO2 sorting algorithm will be compared to established cloud top retrieval algorithms. This will include comparisons with CO2 Slicing and MLEV (Minimum Local Emissivity Variance) applied to SHIS and NAST-I data collected during THORPEX. In addition, the analysis will include a comparison with the Cloud Physics Lidar (CPL) that is capable of independently measuring the cloud top altitude as well as cloud optical depth. From these comparisons the sensitivity of the CO2 sorting algorithm will be investigated.