This is a sociolinguistic study which aims to analyze the types of and functions for code mixing that is used by two Acehnese Scholars in delivering speeches in Tastafi congregations. The researcher is interested in analyzing Islamic speeches because of the Islamic history background in Aceh. This research was conducted by using qualitative method. The data were collected from the speech videos of two Acehnese Islamic scholars, Abu Mudi and Waled Nu. The data in this study were analysed and classified based on type of code mixing proposed by Muysken (2000) and used Hoffman’s (1991) theory to analyze the functions for code mixing in the speeches. The result revealed that insertion is the most frequent type (203 occurences) followed by alternation with 134 occurences, and 83 occurences of lexical congruence. In relation to functions of code mixing, the topic with 247 occurences was found dominantly in the speeches, successively followed by lexical need with 112 occurences, to soften or strengthen the command with 17 occurences, and quoting with 10 occurences. The practical implication from the result that we can put into practice is that when using languages in our communication; it is important to consider the audiences’ native language, and their knowledge of other languages, and take them into account when deciding to use code mixing for better and more effective communication.