While there is ample research on the effect of exposure to foreign language (FL) video materials on developing vocabulary knowledge and listening skills, research on the impact of watching videos on acquiring pronunciation skills, especially in terms of both perception and production, is still in its infancy. This study investigates the effect of viewing an English language video in different subtitling conditions (L1, L2, and no subtitles) on learners' recognition of the correct pronunciation of words and on learners' ability to produce words intelligibly. The relationship between recognition, intelligibility, and comprehension of the video material is also examined. In the study, 54 Polish intermediate learners of English were divided into three experimental groups based on the subtitling condition. They completed pretests and post-tests on recognition and production of words, as well as a comprehension test. The results show that watching the video, either with or without subtitles, significantly improved the recognition of how words are pronounced. As for gains in production, L1 subtitles turned out to be the least effective; however, they led to highest comprehension. These results have practical implications for teachers' use of video material in pronunciation work in the classroom.