Introduction. Mathematics is one of the most important academic subjects in school. However, it is one of the most difficult subjects, and both students and teachers face difficulties in this subject. Teachers have difficulty engaging students in learning the subject. Rarely has the concept of engagement been researched in the Middle East region, in general, and in Oman in particular. Similarly, the concept of self-efficacy has been minimally investigated. The purpose of this study is to test the construct validity of: self-efficacy, engagement, and anxiety in mathematics. Method A multi stage cluster sample (n=900: 420 females; 480 males) was selected from two Omani school directorates. The sample was selected from grades 6, 8 and 10 in Muscat and Aldhahira. Three measures were administered to intact classes. Models of relations among the constructs were developed and tested. Results. The results revealed reasonable construct validity for each construct. Structural equation models indicated that self-efficacy significantly predicted engagement. Neither selfefficacy nor engagement was able to predict anxiety. The results of correlations, SEM as well as dis-criminant analysis, with math anxiety is included, are inconsistent with most previous research. Discussion and Conclusion. For programs and initiatives to promote engagement in mathematics, students should develop confidence in learning mathematics. In addition, any program should assess students' math anxiety and focus more on students with low to medium math anxiety. The nature of math anxiety in Oman raises many questions.