AbstractAccurate volatility forecasting is the key to successful risk analysis. In fact, volatility forecasts lie at the centre of many financial systems, such as value at risk modelling and pricing of derivative securities. This paper is concerned with how to construct stock index volatility predictors using the returns histories of the stocks that define the Index. Specifically, our approach presupposes that the total volatility of the index returns can be explained by the volatility of the related components. Copyright © 2001 John Wiley & Sons, Ltd.