If a university or course of study to apply for accreditation, graduation rate is one of the influential factors. Educational period targeted in 4 years or 8 semesters of study period. But in reality there are still many students who pass beyond the study period. In this case the university or course of study can utilize students' self data to predict the student's graduation rate. One of them by using the concept of data mining. In this research the authors used an a priori algorithm to find the relationship between departments taken at high school level with the level of graduation students. The student's graduation rate is measured by length of study and GPA. The calculation is  using 2 ways that is by manual calculation and by using Tanagra software. Based on the results of the analysis from regular class A students of 2012/2013 which amounted to 23 data it can be found rule, if the majors taken at high school level is SMK, then the possibility of the student will graduate on time with a period of 4 years or less. The GPA that the student will achieve between 3.1 - 3.5. Keywords - Algoritma Apriori, Data mining, Student Graduation, Tanagra