The present study is conducted to explore the relationship among academic achievements, academic emotions and academic ability self-concept of junior middle school students in Mathematics, as well as to provide empirical evidences for a hypothesis of Control-Value Theory of Achievement Emotions proposed by Pekrun（2006）who suggested that academic achievements can produce a reaction on academic emotions through control-related appraisals. A sample of 342 students from a junior middle school ( 49.42% girls; 31.87% grade 7, 30.70% grade 8, 37.43% grade 9) in Hunan Province of China completed Mathematics self-concept questionnaire and Adolescence academic emotions questionnaire. The mathematics achievement indicators were the students’ mathematics test scores in mid-term exam before this investigation. The survey was conducted with the help of the class advisers at the end of April, 2012. The data were used for MANOVA , and Post Hoc Test（LSD）by SPSS, and for SEM by AMOS. The results (See Table 1) of MANOVA and LSD show that in junior middle school: (1) No significant differences exist between positive-high arousal emotions of male students and those of female (p>.05); (2) Positive-low arousal emotions of male students are significantly higher than those of female students((p<.05); (3) Negative-low and negative-high arousal emotions of male students are significantly lower than those of female students (p≦.01); (4) Both positive-high and positive-low arousal emotions of grade 7 students are significantly higher than those of grade 8 and 9 students (p<.01); (5) There are significant variances among negative-low arousal emotions of three different grade students(p<.01) with grade 9 the highest, grade 8 the second and grade 7 the lowest; (6) There are not significant variances among negative-high arousal emotions of three different grade students(p>.05). Taking the influence of gender and grade into control, the results (See Figures 1, 2, 3, 4) of SEM show that: (1) Through academic ability self-concept, mathematic achievements indirectly predict academic emotions with positive predictions on positive emotions and negative predictions on negative emotions; (2) Mathematic achievements have no significant direct effect on academic emotions; (3) All paths are significant at p <.001. The conclusions of this study are: (1) In junior middle school, male students have higher mathematics positive emotions and lower mathematics negative emotions than those of female students; Positive emotions decrease and negative emotions increase with grade. (2) Mathematics academic ability self-concept is the full mediator factor between mathematics academic achievements and mathematics academic emotions. Based on research of Goetz, Frenzel, Hall, & Pekrun in 2008, this article discussed the characteristics of 4 types including 13 mathematics academic emotions and their relationships with mathematics academic achievements and mathematics academic ability self-concept so as to verify the correctness of Pekrun’s assumptions about the relationship between academic achievements and academic emotions.