We study default and endogenous leverage in the laboratory. To this purpose, we develop a general equilibrium model of collateralized borrowing amenable to laboratory implementation and gather experimental data. In the model, leverage is endogenous: agents choose how much to borrow using a risky asset as collateral, and there are no ad-hoc collateral constraints. When the risky asset is financial, namely, its payoff does not depend on ownership (such as a bonds), collateral requirements are high and there is no default. In contrast, when the risky asset is non-financial, namely, its payoff depends on ownership (such as a firm), collateral requirements are lower and default occurs. The experimental outcomes are in line with the theory’s main predictions. The type of collateral, whether financial or not, matters. Default rates and loss from default are higher when the risky asset is non-financial, stemming from laxer collateral requirements. Default rates and collateral requirements are closer to the theoretical predictions as the experiment progresses.