The Cryogenic Infrared Spectrometers and Telescopes for the Atmosphere (CRISTA) experiment aboard the Shuttle Pallet Satellite (SPAS) was successfully flown in early November 1994 (STS 66) and in August 1997 (STS 85). This paper focuses on the first flight of the instrument, which was part of the Atmospheric Laboratory for Application and Science 3 (ATLAS 3) mission of NASA. During a free flying period of 7 days, limb scan measurements of atmospheric infrared emissions were performed in the 4 to 71 μm wavelength region. For improved horizontal resolution, three telescopes (viewing directions) were used that sensed the atmosphere simultaneously. Atmospheric pressures, temperatures, and volume mixing ratios of various trace gases were retrieved from the radiance data by using a fast onion-peeling retrieval technique. This paper gives an overview of the data system including the raw data processing and the temperature and trace gas profile retrieval. Examples of version 1 limb radiance data (level 1 product) and version 1 mixing ratios (level 2 product) of ozone, ClONO2, and CFC-11 are given. A number of important atmospheric transport processes can already be identified in the level 1 limb radiance data. Radiance data of the lower stratosphere (18 km) indicate strong upwelling in some equatorial regions, centered around the Amazon, Congo, and Indonesia. Respective data at the date line are consistent with convection patterns associated with El Ni¿o. Very low CFC-11 mixing ratios occur inside the South Polar vortex and cause low radiance values in a spectral region sensitive to CFC-11 emissions. These low values are a result of considerable downward transport of CFC-11 poor air that occurred during the winter months. Limb radiance profiles and retrieved mixing ratio profiles of CFC-11 indicate downward transport over ~5 km. The accuracy of the retrieved version 1 mixing ratios is rather different for the various trace gases. In the middle atmosphere the estimated systematic error of ozone is ~14%. Ozone data of correlative satellite measurements are well within this error bar. CRISTA agrees, for example, with atmospheric trace molecule spectroscopy sunset measurements typically within 5%. The random error of version 1 ozone mixing ratios is 4%. Similar values apply to other trace gases. These low random errors allow the identification of small and medium scale horizontal and vertical structures in the measured trace gas distributions. Examples of structures in mixing ratio fields of ozone, ClONO2, and CFC-11 are given. ¿ 1999 American Geophysical Union