Changes in promotional expenditure decisions for a brand, as in other marketing decisions, should be based on the expected impact on purchase and consumption behavior as well as on the likely reactions by competitors. Purchase behavior may be predicted from estimated demand functions. Competitive reactions can be anticipated based on estimated reaction functions, which show how changes in an instrument for one brand depend on changes in one or more instruments for other brands. Economic and other arguments suggest that the parameters of such reaction functions (e.g. reaction elasticities) should exhibit systematic patterns. We consider these parameters in an economic framework that also contains two other useful measures, viz. net- and gross demand elasticities. For a defender, the net demand elasticity is the difference between the (positive) demand effect of a reaction (the competitive reaction elasticity multiplied by the corresponding own-brand demand elasticity) and the (negative) cross-brand effect (cross-brand demand elasticity). The gross demand elasticity captures just the (positive) demand effect of the reaction. Net demand-, gross demand- and reaction elasticities are useful yardsticks for managers. Our empirical results suggest that managers' competitive reactions do take into account the consumer response to cross-brand and own-brand activities. For example, the greater a cross-brand elasticity, the greater the corresponding reaction elasticity. Similarly, the greater the own-brand elasticity, the smaller the corresponding reaction elasticity. These effects are consistent with managerial behavior that aims to preserve market shares. We also find that the frequency of changes in other brands' marketing variables has a negative impact on competitive reactions. Interestingly, we do not find that average within-firm competitive reaction elasticities are smaller than for brands offered by different firms. However, we do find that for one firm, this reaction incidence is far lower for sister brands (presumably due to the use of category management), while for another firm, this incidence is higher relative to non-sister brands, consistent with expectations derived from personal interviews of all brand managers. The latter result suggests that the characteristics of reward or resource allocation systems may influence the competitive intensity between managers in the same firm.