In this study it’s studied the prediction modelling of Istanbul Stock Exchange 100 Index (ISE100) with the use of Artificial Neural Network (ANN) Methodology. Also this model has been compared to a linear regression model in order to determine the advantages towards each other. During the study the daily data sets from 24 January 2002 to 24 April 2006 have been used. These datasets have been obtained from the Electronical Data Dissemination System of Central Bank of the Republic of Turkey and the website of the Board of Governors of the Federal Reserve System. The variables which are thought to be able to explain the ISE100 index are determined both from literature researchs and empirical studies. In the first chapters it’s mentioned the stock exchange prediction methods in the framework of Efficient Market Hypothesis and then the ANN methodology is studied in details. Afterwards using of ANN methodology in the field of finance and economics has been discussed. In the last chapters with the using of experience which has been gained from the researchs, a linear regression model and a nonlinear ANN model have been set in order to predict the ISE100 index . In both models the explanatory power of input variables which are foreign exchange, gold price, transaction volume, Fed interest rate and daily dummies have been tested. As a result, it’s seen that an ANN which is modeled with a nonlinear architecture is working more efficiently than a linear regression model in order to explain the ISE100 index. But especially not beening extremely successful to predict the shock times, this model is not appropriate for point estimation.