Personalized recommendation plays an important role on content selection during the adaptive learning process. It is always a challenge on how to recommend effective items to improve learning performance. The aim of this study was to examine the feasibility of applying adaptive testing technology for personalized recommendation. We proposed the adaptation of applicable adaptive testing technology as a solution based on the widely accepted teaching philosophy that providing learners with challenging content can stimulate their potential and improve their performance. Specifically, we adapted item selection algorithms to make the difficulty of recommended items above the learner's current knowledge level on the basis of the uniform scale of learner's ability and item difficulty in the adaptive testing area. Participants were recruited from two classes in a junior middle school, one served as the experimental group by applying the adaptive testing recommendation, and the other served as the control group by applying the random recommendation. The results showed that the experimental group students achieved significantly higher scores and demonstrated higher learning abilities than the control group students. Therefore, the adapted testing technology that is being used for personalized recommendation is effective in improving learning performance.