One of the most remarkable characteristics of collective motion of fish is the emergence of complex migration patterns in which swimming fish are synchronised by remaining together and moving in the same direction. These migration patterns, referred to as fish schools, are often explained using individual based models (IBM’s) that focus on interactions between single individuals. The IBM’s appear to be realistic and robust; however, they are computationally unable to efficiently describe migration of large groups of fish. Here, an approach for developing computationally efficient super-individual based models from simple individual based models for fish migration is proposed. This approach accentuates on ecological mechanisms underlying collective motion of fish, and interaction between them; it explicitly incorporates such important mechanisms in collective motion of fish as fish school splitting and merging.