Abstract: The student‘s education falling is one of the major instructional problems. Among the main tasks and duties of instructional system, persuading the students to learn the cognitive and business skills is one of the most important tasks. So to promote the instructional quality of universities, the educational falling factors must be determined and eliminated. This study is a comparative one has been carry out to investigate the main factors of educational falling in Mathematics students of Guilan Payame Noor University. For this purpose, a population 622 students (302 unsuccessful and 320 successful) ones were selected. The tools that were used of to collect data were Rajers Test for measuring the students self-concept variable Rion Test for measuring IQ level of students variable Mc inerny academic motivation Test for measuring the students academic motivation and the questionnaire that was prepared by the investigator for measuring the other variables. The confidence degree of these tests were 0.80, 0.85 and 0.87 and the confidence degree of the questionnaire was 0.87. This questionnaire was included the questions to assess the students expectance of their future job and the social and personal instructional characteristics. By considering the kind of exist variables the data analysis was conducted by using the t-statistical methods the mann-whitney U Test factor analysis and auditor analysing. The t-test results showed that diploma average IQ level of students academic motivation, self-concept, level of study for students and family income are significant differences between successful and unsuccessful students. Man Whitney Test results showed that the level of education, environment, culture, birth order, according to the professor, students are significant differences between successful and unsuccessful students. Factors affecting students educational falling were fitted using factor analysis. Final model fit revealed that the factor of 11, 5 factors, i.e., low diploma average level, IQ level of student, level of study for students, parents’ education level and low-income parents to explain the causes of educational falling. And finally to predict membership in two groups of successful and unsuccessful students. And drawing a linear model to determine the predictive variables in explaining the dependent variable (academic failure) the regression equation, discriminant analysis was used. The results showed that among 11 existing factors, 6 factors diploma average, IQ level of student, academic motivation, hours of study, parents, education and family income can explaine the students educational falling reasons. And other variables and coefficients for high vilks lambda function were excluded from the equation. In this pattern, the diploma average factor has highest weight in determining the educational falling reasons.