This paper demonstrates a compressed sensing-based single-shot hyperspectral imaging system for combustion diagnostics. The hyperspectral system can capture well-resolved spectra in a 2D plane through a single shot, i.e., converting a 3D data cube of 2D spatial and 1D spectral information into a compressed 2D hyperspectral image. Experimentally, the light emissions are first coded by a random binary pattern to generate the hyperspectral content, which is then sent through a spectrometer. The resulting compressed hyperspectral image is computationally analyzed to recover original 2D spatial and 1D spectral information. C2∗ and CH∗ chemiluminescence emissions of a methane/air flame at various equivalence ratios are measured using the compressed hyperspectral imaging technique. Comparison to traditional measurements shows good agreement in the correlation of emission ratio to equivalence ratio. The technique can be further applied to other laser-based combustion diagnostics.