Digitized periodic material models of size 100(3) nm(3) of mesoporous xerogel Gelsil 50 are reconstructed by use of an evolutionary minimization technique, with two-point probability S-2(r) and volume-based pore size distribution Psi(D) as a hybrid target function. S-2(r) and Psi(D) are derived from small-angle neutron scattering (SANS) and adsorption data, respectively. The nitrogen distribution in Gelsil 50 is characterized in the multilayer adsorption and capillary-condensation regimes by S-2(r) and statistical parameters obtained from in situ SANS data. The fraction of liquid-free pore space phi is calculated from nitrogen adsorption data, and the distribution Psi(corr)(D) of the diameter of the liquid-free pore space at certain relative pressures is derived from Psi(D). The evolutionary algorithm is also used to generate the spatial nitrogen distribution by means of the descriptors phi, S-2(r), and Psi(corr)(D). The morphological parameters obtained from the reconstructs are compared to the respective SANS results.