The cost-effectiveness of policies that provide subsidized health goods is often compromised by the fact that many individuals do not use the goods that are provided to them. Cost-sharing strategies can improve targeting efficiency by inducing self-selection, but have been shown to significantly dampen overall take-up (which is often the primary policy goal). As a potential solution, applying an ordeal mechanism to the distribution of subsidized health goods has been proposed as a way to balance the dual goals of improving targeting efficiency while at the same time maintaining take-up. Another commonly used policy to promote health goods is health training. Despite the work that has been done (separately) on ordeal mechanisms and health training programs, there is a gap in the literature: that is, the effect of ordeal mechanisms without and without health training programs. In this paper, we report the results of a randomized field experiment designed to provide eyeglasses to myopic students of primary schools in rural western China. We test the performance of an ordeal mechanism with and without health training. We emphasize three findings. First, both with and without health training, the ordeal modestly improved targeting efficiency and reduced program costs compared to direct free distribution. Second, in the short run, the training program caused the ordeal mechanism to screen out some individuals who would have used eyeglasses if provided for free. Third, in the medium run, the ordeal mechanism only screened out individuals who would not have used eyeglasses both with the training program and without.