The paper examines the distribution of realized stock return volatility using high-frequency intraday transaction prices from the Dow Jones Industrial Average (DJIA). It finds that the unconditional distributions of realized variances are highly right-skewed, while the logarithmic standard deviations and correlations are approximately normal. Realized volatilities and correlations exhibit strong temporal dependence and are well described by long-memory processes. There is also evidence that realized volatilities and correlations move together in a manner consistent with latent factor structures. The study uses continuously recorded transaction prices to estimate ex post realized daily volatilities by summing squares and cross-products of high-frequency returns. These estimates are model-free and theoretically free from measurement error as the sampling frequency increases. The analysis focuses on 30 DJIA stocks and finds that the results hold for a random sample of 30 other liquid stocks. The study highlights the importance of high-frequency data in capturing the distributional and dynamic properties of stock market volatility. It also shows that the unconditional daily return distributions are leptokurtic, but normalized by realized standard deviations, they are close to normal. The results indicate that the unconditional distributions of realized volatilities and correlations are well-approximated by continuous normal mixtures. The study also finds that the realized volatilities and correlations exhibit strong dynamic dependence and long-memory effects, with the degree of integration around 0.35. The leverage effect, or asymmetry in the relation between past negative and positive returns and future volatilities, is found to be relatively unimportant. The study also finds that the realized correlations exhibit similar asymmetry. The results suggest that the volatility and correlation distributions are well described by a simple factor structure. The study provides a model-free approach to measuring and analyzing stock market volatility, which is important for risk management, portfolio allocation, and asset pricing. The paper also discusses the implications of these findings for the development of improved volatility models and out-of-sample volatility forecasts.The paper examines the distribution of realized stock return volatility using high-frequency intraday transaction prices from the Dow Jones Industrial Average (DJIA). It finds that the unconditional distributions of realized variances are highly right-skewed, while the logarithmic standard deviations and correlations are approximately normal. Realized volatilities and correlations exhibit strong temporal dependence and are well described by long-memory processes. There is also evidence that realized volatilities and correlations move together in a manner consistent with latent factor structures. The study uses continuously recorded transaction prices to estimate ex post realized daily volatilities by summing squares and cross-products of high-frequency returns. These estimates are model-free and theoretically free from measurement error as the sampling frequency increases. The analysis focuses on 30 DJIA stocks and finds that the results hold for a random sample of 30 other liquid stocks. The study highlights the importance of high-frequency data in capturing the distributional and dynamic properties of stock market volatility. It also shows that the unconditional daily return distributions are leptokurtic, but normalized by realized standard deviations, they are close to normal. The results indicate that the unconditional distributions of realized volatilities and correlations are well-approximated by continuous normal mixtures. The study also finds that the realized volatilities and correlations exhibit strong dynamic dependence and long-memory effects, with the degree of integration around 0.35. The leverage effect, or asymmetry in the relation between past negative and positive returns and future volatilities, is found to be relatively unimportant. The study also finds that the realized correlations exhibit similar asymmetry. The results suggest that the volatility and correlation distributions are well described by a simple factor structure. The study provides a model-free approach to measuring and analyzing stock market volatility, which is important for risk management, portfolio allocation, and asset pricing. The paper also discusses the implications of these findings for the development of improved volatility models and out-of-sample volatility forecasts.