We examine theories of leverage and debt maturity, focusing on the impact of firms' investment opportunity sets and regulatory environments in determining these policies. Using results on strategic complementarities, we identify sufficient conditions for the theory to have testable implications for reduced-form and structural-equation regression coefficients. Obtaining testable implications for structural equations requires less from the theory but more from the data than the reduced-form specification because it requires an instrumental-variables approach. We examine this trade-off between theory and statistical methods and provide tests using two decades of data for over 5000 industrial firms. © 2002 Elsevier Science B.V. All rights reserved.