This article describes a study in which the administration of two health surveys, the Veterans SF-36 and SF-12, by telephone and mail-out was used to assess the differences in the health surveys' costs and scores by mode of administration and determine which mode was cost-efficient. The study employed a crossover design: after 12 unsuccessful attempts to contact patients by telephone, the patients were administered the survey by mail, and after 2 unsuccessful mail-outs, up to three attempts were made to interview the patients by telephone. The analysis of the data showed that mail administration, with or without crossover to telephone, was more cost-efficient than telephone administration, having both lower average total and variable costs per completed questionnaire. Overall, telephone administration was about 30% more expensive that mail administration, primarily due to the cost of labor. The marginal cost of an additional completed Veterans SF-12 or Veterans SF-36 was also substantially lower for mail administration. Mail administration without crossover to telephone administration was the most cost-efficient strategy for administering both the Veterans SF-12 and SF-36. The results of this study strongly suggest the need to consider the mode of administration if questionnaires like the Veterans SF-12 or SF-36 are to be used to assess health outcomes within and across large health care systems.