The results of infrared absorption experiments, performed on a steady, nonreacting flow of methane and argon to test a new inversion method for tomographic reconstruction of spatially nonuniform distributions, are presented. The absorption coefficient distributions to be reconstructed are expressed as a weighted sum of Karhunen-Loève eigenfunctions, formed by applying the Karhunen-Loève procedure to a training set containing aprioriinformation regarding the distributions to be reconstructed. Reconstruction of asymmetric absorption coefficient distributions is accomplished by utilizing a set of laser absorption measurements obtained using an infrared helium neon laser. Issues pertinent to the practical application of the new inversion method, such as the construction of a training set using probe sampling, the effect of experimental error, and the precise modeling of laser absoroption paths, are discussed. Reconstruction of absorption coefficient distributions from only 42 laser absorption measurements typically achieve normalized errors of 10% or less over 80% of the domain.