"The proposal aims at the development of a generally applicable semiempirical approach that goes beyond the current standard model by including explicit orthogonalization and dispersion terms into the semiempirical Hamiltonian. We have recently shown in preliminary work on organic molecules that such orthogonalization models (OMx = OM1, OM2, OM3) are significantly more accurate than standard semiempirical methods (AM1, PM3, PM6) both for ground-state and excited-state properties, at comparable computational costs. We plan to improve the OMx models by incorporating dispersion corrections (OMx-D) and by extending the formalism from an sp to an spd basis (OMx-DE). The resulting approaches will be parameterized for all chemically important main-group elements and transition metals to generate the next generation of generally applicable semiempirical methods. These methods are designed to fill the currently existing gap between density functional theory (DFT) and classical force field approaches. Being about 1,000 times faster than DFT, and being capable of treating electronic events (unlike classical force fields), OMx-based methods are expected to enable realistic electronic structure calculations, with useful accuracy, on large complex systems in all branches of chemistry. Especially when applied in a multi-method strategy, with synergistic use of different computational tools, this will allow the modelling of many chemically relevant systems that are currently beyond reach for computational chemistry. Proof-of-concept applications will address the reaction mechanisms of enzymatic reactions (biocatalysis) and electronically excited states (organic solar cells, photoactive proteins, excited-state dynamics in complex systems). The successful development of generally applicable OMx-based methods will provide a breakthrough in computational chemistry by opening up new areas of application."