This study analyzed the use of code-mixing in the on-air promotion of GEN 98.7 FM. It aimed to answer two research questions, which are: what codes occur in GEN 98.7 FM’s on- air promotions, and what types of code-mixing are used based on Muysken’s (2000) theory. Since radio relies on spoken language, announcers often use code-mixing to make speech more natural and communicative. This study focused on GEN FM on-air promotion. A qualitative approach was used in this study. Data were collected from broadcasts on February 14, 2026, resulting in 99 sentences. Indonesian was treated as the matrix language and identified using KBBI Online (Badan Pengembangan dan Pembinaan Bahasa, n.d.), while the other codes were identified using relevant dictionaries. The findings showed that English, Betawi, Javanese, and Sundanese occurred in GEN 98.7 FM on-air promotion. In addition, the data also contained words that had undergone word formation process. This study only found two out of three types of code-mixing, which are insertion and alternation types, with insertion being the most frequent type. Congruent lexicalization did not appear.