This article examines the methodological implications of the societal perspective in cost-effectiveness analyses in which the costs of health-care interventions are defined as the sums of direct and indirect costs. In the model of cost-effectiveness analysis in which the planner distributes patients among many treatments for many illnesses, the definition requires that total indirect costs be constrained, and the article proposes an iterative computational procedure for choosing a constraint under the assumption that the planner maintains a target trade-off rate between losses of health benefits and reductions in indirect costs. In the more common model in which the planner decides which of two treatments for one illness to provide to patients, the adoption of a societal perspective introduces ambiguities into the welfare properties of the decision rule, and in general the conclusion that a treatment is cost-effective is valid only if switching or assigning patients to the cost-effective treatment does not increase the planner's total direct costs.