Recent improvements in spectrum excitation, recording, and processing capabilities have led to enormous enhancement in the quality and quantity of spectroscopic data sets. We describe here a pattern recognition technique, extended cross correlation (XCC), that is well suited to take advantage of large, high quality data sets. In particular, spectra are used to decode each other without any knowledge of or assumptions about the patterns that are sought. This paper describes the motivation for and construction of the XCC, and illustrates one of its simplest applications: To identify, in spectra of mixtures of chemical species, which peaks correspond to which chemical species. This application of the XCC is illustrated with both synthetic data and experimental data on mixtures of ammonia isotopic species. (C) 1997 American Institute of Physics. [S0021-9606(97)01944-2].