This study aims at exploring the potential of extensive TV viewing for L2 vocabulary learning, and the effects associated with the language of the on-screen text (L1 or L2), type of instruction (pre-teaching target items or not) and learners' proficiency. A total of 106 secondary school students (Grade 8) divided into 4 classes participated in a one-year pedagogical intervention, viewing 24 episodes of a TV series under four experimental conditions with each class being assigned to a different treatment: (1) captions and pre-teaching, (2) captions and non-pre-teaching, (3) subtitles and pre-teaching, and (4) subtitles and non-pre-teaching. Following a pre-/post-test design, form recall and meaning recall gains were examined. Results showed that participants learnt vocabulary in all four conditions, with greater gains in recalling form than in recalling form and meaning. The analysis also showed that, overall, groups that were pre-taught the target items performed better, independently of the language of the on-screen text. An important finding is the role of learners' proficiency prior to the intervention, with higher proficiency related to higher gains. The study contributes to the area of foreign language learning through audio-visual input with results from a longitudinal, classroom-based study with adolescent learners.