Modeling of material behavior generally involves the development of a mathematical model derived from observations and experimental data. An alternative way discussed in this paper, is neural network-based modeling that is a subfield of artificial intelligence. The main benefit in using a neural network approach is that the network is built directly from experimental data using the self-organising capabilities of the neural network. In this paper, size effects in fracture of cementitious materials are modeled with a back-propagation neural network. The results of neural network-based size effect law look viable and very promising. Copyright (C) 1996 Elsevier Science Ltd.