The size and conformational flexibility of biomolecules calls for computationally efficient quantum chemical approaches capable of treating ground-state and excited state properties of proteins, DNA and their complexes. The speed and versatility of semi-empirical methods make them very useful for modeling biomolecules. Development and application of semi-empirical schemes to explore the properties and functions of the complex system is considered. Specific strengths and weaknesses of the methods are highlighted. Extension of the methods to transition metals and their further development are considered.