This study examines the use of code mixing on the YouTube channels of beauty vloggers Tasya Farasya and Nanda Arsyinta. It seeks to examine the various forms of code mixing according to Muysken's (2000) theory as well as the fundamental causes of the two beauty vloggers' language mixing in their videos according to Saville-Troike's (1986) theory. This study uses qualitative research methodology. The results of the research show that 210 extracts describe the types of code mixing found in the utterances of the two beauty vloggers Nanda Arsyinta and Tasya Farasya on YouTube channel. Based on Muysken's theory (2000), there are 3 types of reasons were also found underlying speakers mixing their languages, namely insertion (159), alternation (6), and congruent lexicalization (45). Based on Saville-Troike's theory (1986), 3 types of reasons were also found that underlying speakers' code-mixing, namely to strengthen requests or commands (5), lexical need (84), and to exclude other people (54). Additionally, the researcher found one reason that was not included in the Saville-Troike theory because the researcher felt that the three reasons put forward by Saville-Troike were not enough to cover all the extracts found. Apart from that, to strengthen the new reasons found, the researcher watched other videos from the two beauty vloggers and then concluded that because of habits are one of the appropriate reasons to use as a basis for speakers to mix their languages. There are 67 extracts found in this case. Based on 210 data observed, the research concluded that the most dominant type of code-mixing was insertion, while the most dominant reason why speakers mixed their languages was lexical need.Keywords: Code-Mixing, Types, Reasons, Beauty Vloggers, YouTube Videos