The purpose of the study was to determine the number of latent classes in the performance of eighth-grade students on the National Test for the quality of education in the fields of science, then to reveal the probability of answering each item of the science test by students across the different latent class levels, and then to study the extent of agreement between Latent Class levels and performance levels approved by the Ministry of Education. To achieve the aim of the study, the researcher analyzed the response patterns of eighth grade students on the and science test using the Mplus software. The study sample, on which statistical processing was applied, consisted of 3000 male and female eighth grade students in Jordan. The results of the study showed that there are three latent classes for the national science test based on students’ response patterns. The probability of students answering the test items played an important role in differentiating between students’ abilities in the form of latent classes and supported the three latent levels model. Finally, using the Chi-square test and contingecy coefficient, the study concluded that there is agreement between the latent performance levels and the performance levels approved by the Ministry of Education. The study concluded with a set of recommendations that there is a clear weakness in performance in the science test, and the reasons for such weakness are unknown, which necessitates the need to investigate whether this weakness reflects an actual weakness in the subject or not, and to reconsider the cut-off scores adopted by the Ministry of Education and adopt one of the known methods in determining cut-off scores instead of relying on the arbitrary method. The four levels used by at the Ministry of Education are not distinct in classifying students, as the basic and partial mastery levels can be combined together.