This study reports the interview responses of ten electronic resource selectors at the University of North Carolina at Chapel Hill. The interviews were conducted to determine how selectors analyze usage statistics and whether selectors would prefer to use less specific, more specific, or a combination of both kinds of statistics to evaluate electronic indexes and databases. Another goal was to determine whether selectors would prefer to use citation analysis, vendor-generated statistics, or a combination of these methods to evaluate electronic journals. Most participants consider statistics within the context of the cost and the content of the resources. Most selectors would prefer to use a combination of less and more specific measures for evaluating electronic indexes and databases and a combination of citation analysis and vendor-generated statistics for evaluating electronic journals. Selectors' rankings of individual measures varied considerably according to their judgments of which measures indicated the highest level of use.