With the rise of computer-assisted online learning tools such as the "Language Reactor," advanced captioning options such as dual subtitles (i.e. simultaneous display of on-screen texts in the first language and the second/foreign language) have gained much attention. However, little is known about how dual-subtitled viewing may be optimised to foster second/foreign language learning in multimedia learning environments. In this study, I extended Laufer and Hulstijn ("Applied Linguistics," 22(1), 1-26, 2001) Involvement Load Hypothesis (ILH) to the context of audiovisual input and examined the effects of dual-subtitled viewing under multiple ILH-based interventions on vocabulary learning. University Chinese-speaking low-intermediate EFL learners (n = 72) were divided into four groups and exposed to dual-subtitled videos with varying ILH-based interventions. Using a counterbalanced 4 × 4 experimental design, I compared learners' immediate vocabulary gains and retention of target words under four different treatments across four experimental sessions, including interventions featuring an involvement load of 4 (moderate "need" + "search" + strong "evaluation"), an involvement load of 3 (moderate "need" + "search" + moderate "evaluation"), an involvement load of 2 (moderate "need" + "search"), and an involvement load of 0 (absence of "need," "search" or "evaluation"). The approach to data analysis was twofold: between-groups and within-groups. ANOVA results revealed that vocabulary gains were the highest when learners viewed dual-subtitled videos under treatments with higher involvement loads. The article ends with a consideration of practical and theoretical implications for language learning through this form of audiovisual multimedia learning condition.