This study investigates the impact of a high-level and a low-level process-based L2 comprehension training on L1 French students' listening performances. These students (N = 108) of English, Spanish, and German came from five different classes. After a L2 comprehension pretest, the participants were divided into two experimental groups. During the experimental phase, the two groups listened to the same three documents in their respective L2. The first group undertook listening comprehension activities relying on low-level processes, while the second group undertook activities aiming at fostering their use of high-level processes. These second types of activities were based on the Metacognitive Awareness Listening Questionnaire (MALQ) (Vandergrift, Goh, Mareschal & Tafaghodatari, 2006). As hypothesized, less skilled listeners trained to automatize low-level processes, scored significantly higher on the comprehension posttest than on the pretest. However, the hypothesis that more skilled listeners trained to develop high-level listening processes, would score better on the posttest than on the pretest could not be verified. Nevertheless, we observed some tendencies showing that more skilled learners had drawn some benefits from the training. In our conclusion, we argue that the impact of a high-level and a low-level process-based L2 comprehension training strongly depends on learners' initial levels.