We develop and structurally estimate a dynamic model of corporate liquidity and risk management. When external finance is costly, liquid funds provide corporations with instruments to absorb and react to shocks. Making optimal use of liquid funds means transferring them to times and states where they are most valuable. In the model, firms can transfer liquidity across time using cash and across states drawing on credit lines subject to debt capacity constraints. Optimal liquidity management arises as a trade-off between conditional liquidity with credit lines subject to collateral constraints and uncontingent liquidity using cash. The estimated model explains well the cross-sectional and time series patterns of corporate liquidity management: Small and constrained firms use cash to provide liquidity to fund investment opportunities, and large and unconstrained firms rely on credit lines. While equity issuances are used to replenish cash balances, credit lines fund unanticipated investment opportunities. To solve the model, we develop a novel and efficient approach to dynamic programming relying on linear programming, that is more widely applicable to high-dimensional dynamic models.