The need for estimating Available Transfer Capability (ATC) has been well understood in the deregulated environment as proper estimation of its value ensures the economically efficient operation of restructured power systems. This paper proposes a method to estimate ATC of selected interface as a function of power flow through various branches. The conventional methods estimate ATC as a function of real power loads, generation scheduling and outages. The power flows through various branches are influenced by the above said factors. ATC can also be estimated as a function of power flow through various branches. The feed forward back propagation algorithm of Artificial Neural Networks has been used to develop the ATC estimator. Only the branch power flows that are having considerable effect on the selected branch power flow are considered. IEEE- 30 bus and AEP-118 bus system are used to demonstrate the feasibility of the proposed model.