This paper presents a structural model of intraday price formation that incorporates both public information shocks and microstructure effects. The model helps explain observed patterns in bid-ask spreads, price volatility, transaction costs, and autocorrelations of transaction returns and quote revisions. For example, it explains why the variance of transaction price changes is U-shaped while the variance of ask price changes declines, why the bid-ask spread is U-shaped despite decreasing information asymmetry, and why transaction price changes have large negative autocorrelations while ask price changes have small negative autocorrelations. The model's parameters also provide a natural metric for price discovery and effective trading costs. The paper develops and estimates a structural model of price formation that captures many microstructure frictions in a unified setting. The model incorporates public information shocks and microstructure effects, including the possibility of crosses within the quoted bid-ask spread, autocorrelation of order flow, and trading frictions from asymmetric information, dealer costs, and price discreteness. The model links two important areas of the literature: the sources of intraday price volatility and the measurement of bid-ask spread components. The model is estimated using transaction-level data for NYSE-listed stocks. The parameters are estimated in a setting with high comfort level in terms of estimation, though not necessarily model error. The model's parameters are interpreted in terms of quote data and transaction price changes-based moment restrictions not used in estimation. The model's results show that both information flows and trading frictions are important in explaining intraday price volatility. Information asymmetry decreases throughout the day, consistent with theoretical models and experimental evidence. Dealer costs increase over the day, leading to a U-shaped bid-ask spread. The model provides an estimator of execution costs that accounts for the possibility of orders executing within the bid-ask spread, as well as information and inventory effects. The cost of transacting is significantly smaller than the bid-ask spread once the probability of executing within the quotes is considered. This measure of execution costs increases over the day, consistent with concentrated trading at the open by discretionary liquidity traders. The model provides insights into the determinants of the autocorrelations of quotes and returns, as well as other moments such as the variance of quote changes. The model's results closely resemble the actual autocorrelations of the data. The paper concludes that the model provides a useful framework for understanding intraday price movements and the factors influencing them.This paper presents a structural model of intraday price formation that incorporates both public information shocks and microstructure effects. The model helps explain observed patterns in bid-ask spreads, price volatility, transaction costs, and autocorrelations of transaction returns and quote revisions. For example, it explains why the variance of transaction price changes is U-shaped while the variance of ask price changes declines, why the bid-ask spread is U-shaped despite decreasing information asymmetry, and why transaction price changes have large negative autocorrelations while ask price changes have small negative autocorrelations. The model's parameters also provide a natural metric for price discovery and effective trading costs. The paper develops and estimates a structural model of price formation that captures many microstructure frictions in a unified setting. The model incorporates public information shocks and microstructure effects, including the possibility of crosses within the quoted bid-ask spread, autocorrelation of order flow, and trading frictions from asymmetric information, dealer costs, and price discreteness. The model links two important areas of the literature: the sources of intraday price volatility and the measurement of bid-ask spread components. The model is estimated using transaction-level data for NYSE-listed stocks. The parameters are estimated in a setting with high comfort level in terms of estimation, though not necessarily model error. The model's parameters are interpreted in terms of quote data and transaction price changes-based moment restrictions not used in estimation. The model's results show that both information flows and trading frictions are important in explaining intraday price volatility. Information asymmetry decreases throughout the day, consistent with theoretical models and experimental evidence. Dealer costs increase over the day, leading to a U-shaped bid-ask spread. The model provides an estimator of execution costs that accounts for the possibility of orders executing within the bid-ask spread, as well as information and inventory effects. The cost of transacting is significantly smaller than the bid-ask spread once the probability of executing within the quotes is considered. This measure of execution costs increases over the day, consistent with concentrated trading at the open by discretionary liquidity traders. The model provides insights into the determinants of the autocorrelations of quotes and returns, as well as other moments such as the variance of quote changes. The model's results closely resemble the actual autocorrelations of the data. The paper concludes that the model provides a useful framework for understanding intraday price movements and the factors influencing them.