The purpose of this study was to analyse the effect of emotional intelligence between the perception of autonomy support from physical education teachers and self-esteem (positive and negative) in secondary-school physical education students. The study design was observational, descriptive, and cross-sectional. In total, 1069 secondary-school physical education students participated (Mage = 14.55; SD = 1.54) (51.2% female; 48.8% male). The following scales were used: The Learning Climate Questionnaire adapted to Physical Education (i.e., autonomy support), the Trait Meta-Mood Scale-24 (i.e., emotional attention, emotional clarity, and emotional repair), and the Rosenberg Self-Esteem Scale (positive self-esteem, negative self-esteem). A structural equation model was performed with the latent variables controlled by age, sex, and the students’ educational centre. The main results indicate that the explained variance was 37% for positive self-esteem and 26% for negative self-esteem. In addition, autonomy support directly predicts emotional intelligence (p < 0.05) and positive self-esteem (p < 0.001). On the other hand, all indirect effects of autonomy support on self-esteem across emotional intelligence were significant at p < 0.001. Finally, emotional clarity and emotional repair had a mediating effect on self-esteem, and it improves the total effect of autonomy support on positive self-esteem with values of β = 0.14 and β = 0.19, respectively, and a value of β = −0.07 and β = −0.06 for negative self-esteem. The findings reveal the necessity to improve emotional clarity and emotional repair in secondary-school students in improving positive self-esteem through the perception of autonomy support from the physical education teacher.