Numerical Monte Carlo simulations were conducted to assess dispersion of reactive solutes in two-dimensional physically and chemically heterogeneous porous media, using random fields with assigned correlation structure for hydraulic conductivity and liner adsorption coefficient. Conditions under which linearization of adsorption is valid are discussed. Lognormal distributions of hydraulic conductivity and adsorption coefficient were assumed. Calculations have been performed for positive and negative correlation between hydraulic conductivity and adsorption coefficient, and for uncorrelated cases. Effects of varying different properties including mean and average adsorption coefficient, physical and chemical integral scale, and variance of hydraulic conductivity on dispersive behavior are shown. A larger mean sorption coefficient enhances plume spreading in uncorrelated and in negatively correlated cases. In positively correlated cases, counteracting effects of physical and chemical heterogeneity play an important role. The outcome of these counteracting effects depends on the mean, variance, and integral scales of the spatially variable properties. The analytical solutions, derived in paper 1 (Bellin et al., this issue), reveal a good agreement with the numerical results in a significant range of heterogeneities. The generally surprisingly good agreement of the analytical solutions with the numerically obtained results can possibly be attributed to opposing effects of nonlinearities neglected in the derivation of the analytical solutions. In the case of strong physical heterogeneity the analytical solutions perform slightly better than in the case of strong chemical heterogeneity. ¿ American Geophysical Union 1993