In the past few decades considerable effort has been expended in characterizing and modeling financial time series. A number of stylized facts have been identified, and volatility clustering or the tendency toward persistence has emerged as the central feature. In this paper we propose an appropriately defined conditional probability as a new measure of volatility clustering. We test this measure by applying it to different stock market data, and we uncover a rich temporal structure in volatility fluctuations described very well by a scaling relation. The scale factor used in the scaling provides a direct measure of volatility clustering; such a measure may be used for developing techniques for option pricing, risk management, and economic forecasting. In addition, we present a stochastic volatility model that can display many of the salient features exhibited by volatilities of empirical financial time series, including the behavior of conditional probabilities that we have deduced.