A modified adaptive algebraic reconstruction technique (MAART) with an auto-adjustment relaxation parameter and smoothness regularization is developed to reveal the tomographic reconstruction of H2O distribution in combustion from incomplete projections. Determinations of relaxation parameter and regularization factor are discussed with regard to the consideration of improvement in reconstruction and reduction of computational burden. A two-wavelength scheme from tunable diode laser absorption measurement, 7444.36 and 7185.59 cm ?1, is used to simplify the nonlinear solution problem for obtaining the tomographic distributions of concentration and temperature simultaneously. Results of calculations demonstrate that MAART can perform the reconstruction results more efficiently with little complex modification and much lower iterations as compared with the traditional algebraic reconstruction technique (ART) or simultaneous iterative reconstruction technique (SIRT) methods. The stability of the algorithm is studied by reconstruction from projections with random noise at different levels to demonstrate the dependence of reconstruction results on precise line-of-sight measurements.