There is an ongoing transition in education from paper-based learning and testing to digital learning and testing. The purpose of the present research was to examine whether the relative effectiveness of digital and paper-based learning depends on the medium of testing in the context of foreign-language vocabulary learning. In a controlled experiment, young adults (N = 79) studied and practiced novel foreign-language vocabulary words using two study methods (restudying or retrieval practice) and were then tested on their memory for these words to assess learning. The study medium and the test medium were either congruent (i.e., paper-based learning and testing; digital learning and testing) or incongruent (paper-based learning and digital testing; digital learning and paper-based testing). The results revealed a study-test medium congruency effect: Paper-based learning yielded better test performance than digital learning when the test was conducted on paper, but this effect was eliminated when the test was digital. This effect may have important practical educational implications as it challenges common practices for vocabulary learning such as using digital tools to study vocabulary for on-paper memory tests.