The objective of this paper is to examine the dynamic interrelations among major world stock markets through the use of artificial neural networks. The data was derived from daily stock market indices of the major world stock markets of Canada, France, Germany, Japan, United Kingdom (UK), the United States (US), and the world excluding US (World). Multilayer Perceptron models with logistic activation functions were better able to foresee the daily stock returns than the traditional forecasting models, in terms of lower mean squared errors. Furthermore, a multilayer perceptron with five units in the hidden layer seemed to predict more precisely the returns of stock indices than a neural network with two hidden elements. Hence, it is inferred that neural systems could be used as an alternative or supplemental method for predicting financial variables and thus justified the potential use of these model by practitioners.