Usage statistics for electronic resources are needed, and highly desirable, for many reasons. It is encouraging to see the beginnings of quality, reliable usage data. This data can form the basis of economic decisions (selection and cancellation) that make a great deal of sense in the context of the individual library. However, the cumulative effects of such decisions could have serious implications for scholarly communications. For example, the journals of small research communities could easily be vulnerable to mass cancellations, and might fold. Fortunately, open access provides an alternative. The question of whether the impact of local decisions on scholarly communications as a whole should be taken into account in collection development policies is raised. The possibility that usage statistics could form the basis for a usage-based pricing system is discussed, and found to be highly inadvisable, as usage-based pricing tends to discourage usage.