This study is about the development of an Artificial Neural Network (ANN) based practical rapid assessment method for Reinforced Concrete (RC) buildings by using the minimum possible number of input data. The problem is formulated as a classification problem and evaluated as two sub-problems. Feed Forward Back Propagation (FFBP) and Generalized Regression Neural Networks (GRNNs) are used in each case and eight different ANN models are developed. To develop ANN models, a total of 402 residential building models are generated of three types and up to eight storeys. The earthquake performance of these building models is investigated through the nonlinear incremental mode combination method. By using the building properties as inputs and the results of structural analyses as outputs, the ANN models are trained and tested. Additionally, existing buildings are used for validation. The results show that the earthquake behavior of RC buildings can be predicted successfully using an ANN.