The theoretical case for industrialpolicy is a strong one. The market failures that industrialpolicies target in markets for credit, labor, products, andknowledge have long been at the core of what developmenteconomists study. The conventional case against industrialpolicy rests on practical difficulties with itsimplementation. Even though the issues could in principle besettled by empirical evidence, the evidence to date remainsuninformative. Moreover, the conceptual difficultiesinvolved in statistical inference in this area are so greatthat it is hard to see how statistical evidence could everyield a convincing verdict. A review of industrial policy inthree non-Asian settings El Salvador, Uruguay, and SouthAfrica highlights the extensive amount of industrial policythat is already being carried out and frames the need forindustrial policy in the specific circumstances ofindividual countries. The traditional informational andbureaucratic constraints on the exercise of industrialpolicy are not givens; they can be molded and rendered lessbinding through appropriate institutional design. Three keydesign attributes that industrial policy must possess areembeddedness, carrots-and-sticks, and accountability.