In this article, we propose a stochastic search-based method, namely genetic algorithm (GA) and simulated annealing (SA) in conjunction with density functional theory (DFT) to evaluate global and local minimum structures of (TiO2)n clusters with n = 1–12. Once the structures are established, we evaluate the infrared spectroscopic modes, cluster formation energy, vertical excitation energy, vertical ionization potential, vertical electron affinity, highest occupied molecular orbital (HOMO)-lowest unoccupied molecular orbital (LUMO) gaps, and so forth. We show that an initial determination of structure using stochastic techniques (GA/SA), also popularly known as natural algorithms as their working principle mimics certain natural processes, and following it up with density functional calculations lead to high-quality structures for these systems. We have shown that the clusters tend to form three-dimensional networks. We compare our results with the available experimental and theoretical results. The results obtained from SA/GA-DFT technique agree well with available theoretical and experimental data of literature. © 2013 Wiley Periodicals, Inc.