Positioned in the Hong Kong education context, this article evaluates the effects of teaching science through home and second languages (i.e. Chinese and English) in Secondary 2 (or eighth grade) science classrooms. A total of 479 students, divided into two language instruction groups, participated in a teaching intervention comprising 16 lessons on the topic of 'Making Use of Electricity'. Informed by the results of a mixed-methods study with a quasi-experimental design, with data collected from science diagnostic tests, inquiry questions and focus group interviews, this article reports that Chinese is the most advantageous language of instruction for low- and middle-ability science students, whereas English is more favourable for their high-achieving peers. Whilst Hong Kong students who learnt the focal topic in English were able to rid themselves of certain naive ideas generated from the translation of science terms into Chinese (e.g. pencil '[Chinese characters omitted]' is translated as 'lead pen' in Chinese), they were also found to have misconceptions about certain scientific concepts. For example, they were confused about 'open' and 'closed' circuits because they mixed up the words 'open' ('[Chinese characters omitted]') and 'switch on' ('[Chinese characters omitted]') in English. The study's broad implications for language support and a mixed-code approach in science teaching worldwide are discussed.