The aim of this research is to determine classification in which the level of accuracy in Turkish universities rankings is detected by the international assessments according to the independent variables PhD students ratio, the number of students per faculty member and the article scores. The data of research were obtained from University Ranking by Academic Performance (URAP) in Turkey. The data were divided into three groups to international university ranking the top 500 (group A), 500 to 1000 (group B), and 1000 and 1500 (group C). Discriminant analysis was used to analyze the data. According to the findings of the study "PhD students' ratio" is variable to distinguish groups of universities who have contributed the most variable. Then "article score" and the last "the number of students per faculty member" are seen. Classification results are analyzed, a total of 10 universities in the A group, 9 (90%), a total of 15 universities in the B group, 10 (66.7%) and, finally, a total of 26 universities in the C group and 19 (73.1%) were classified correctly. The total percentage of correct classification of discriminant function is 74.5%. Classification accuracy of the discriminant analysis is higher than chance criterion (33%). In other words, function obtained from the research makes a more accurate classification than change classification.