An important issue in modern solid state chemistry is the development of a general methodology to predict the possible (meta)-stable modifications of a solid. This requires the global exploration of the energy landscape of the chemical system, since each stable phase corresponds to a locally ergodic region of the landscape. The global search in the lead sulfide system has been performed with simulated annealing on the ab initio level, while zinc oxide was studied with an empirical potential using simulated annealing, both at standard and elevated pressure (up to 100 GPa). The local optimization of the modifications found in the PbS system was performed using various density functionals. Next, the energy E(V ) and enthalpy H(p) as function of volume and pressure, respectively, were computed for these modifications and their electronic structure was analyzed. The structures found for ZnO were locally optimized on ab initio level (DFT and Hartree-Fock). In both systems the structures found were in good agreement with the experiment. Furthermore, we employed the threshold algorithm to explore the barrier structure of the landscape of ZnO as function of the number of formula units in the simulation cell. Based on the barrier and minima information 2-D models of the energy landscape were constructed.