The paper presents an econometric analysis of realized volatility and its use in estimating stochastic volatility (SV) models. Realized volatility, derived from intraday price data, is shown to be a useful measure for studying the stochastic properties of financial returns. Under a general SV model, the paper derives the moments and asymptotic distribution of the realized volatility error—defined as the difference between realized volatility and the discretized integrated volatility (actual volatility). These properties enable the estimation of SV model parameters without the need for simulation-intensive methods. Key findings include the derivation of the asymptotic distribution of the realized volatility error scaled by the square root of the number of intraday observations, which is shown to be independent of the drift and risk premium parameters. The paper also discusses the second-order properties of realized volatility and its relationship to actual volatility, showing that realized volatility can be used to estimate actual volatility when the drift and risk premium are zero. The paper further explores the use of model-based estimators of actual volatility, which can provide more accurate estimates than model-free methods, especially when the number of intraday observations is large. Empirical examples illustrate the practical implications of these findings, showing that realized volatility can be used to estimate actual volatility with reasonable accuracy, although the presence of biases must be considered. The paper also discusses the asymptotic distribution of the realized volatility error, showing that it follows a normal variance mixture distribution, and highlights the importance of considering the level of volatility when estimating the variance of the realized volatility error. The paper concludes that model-based approaches can significantly reduce the mean-square error in estimating actual volatility, and that the use of realized volatility in estimating SV models can provide valuable insights into the stochastic properties of financial returns. The results are supported by simulations and empirical illustrations, demonstrating the effectiveness of realized volatility in estimating SV models.The paper presents an econometric analysis of realized volatility and its use in estimating stochastic volatility (SV) models. Realized volatility, derived from intraday price data, is shown to be a useful measure for studying the stochastic properties of financial returns. Under a general SV model, the paper derives the moments and asymptotic distribution of the realized volatility error—defined as the difference between realized volatility and the discretized integrated volatility (actual volatility). These properties enable the estimation of SV model parameters without the need for simulation-intensive methods. Key findings include the derivation of the asymptotic distribution of the realized volatility error scaled by the square root of the number of intraday observations, which is shown to be independent of the drift and risk premium parameters. The paper also discusses the second-order properties of realized volatility and its relationship to actual volatility, showing that realized volatility can be used to estimate actual volatility when the drift and risk premium are zero. The paper further explores the use of model-based estimators of actual volatility, which can provide more accurate estimates than model-free methods, especially when the number of intraday observations is large. Empirical examples illustrate the practical implications of these findings, showing that realized volatility can be used to estimate actual volatility with reasonable accuracy, although the presence of biases must be considered. The paper also discusses the asymptotic distribution of the realized volatility error, showing that it follows a normal variance mixture distribution, and highlights the importance of considering the level of volatility when estimating the variance of the realized volatility error. The paper concludes that model-based approaches can significantly reduce the mean-square error in estimating actual volatility, and that the use of realized volatility in estimating SV models can provide valuable insights into the stochastic properties of financial returns. The results are supported by simulations and empirical illustrations, demonstrating the effectiveness of realized volatility in estimating SV models.