The concept of imaging spectrometry, or hyperspectral imaging, is becoming increasingly popular in scientific communities in recent years. Hyperspectral imaging data covering the spectral region between 0.4 and 2.5 Am and collected from aircraft and satellite platforms have been used in the study of the earth's atmosphere, land surface, and ocean color properties, as well as on planetary missions. In order to make such quantitative studies, accurate radiometric and spectral calibrations of hyperspectral imaging data are necessary. Calibration coefficients for all detectors in an imaging spectrometer obtained in a laboratory may need to be adjusted when applied to data obtained from an aircraft or a satellite platform. Shifts in channel center wavelengths and changes in spectral resolution may occur when an instrument is airborne or spaceborne due to vibrations, and to changes in instrument temperature and pressure. In this paper, we describe an algorithm for refining spectral calibrations of imaging spectrometer data using observed features in the data itself. The algorithm is based on spectrummatching of atmospheric water vapor, oxygen, and carbon dioxide bands, and solar Fraunhofer lines. It has been applied to real data sets acquired with airborne and spaceborne imaging spectrometers.