Purpose: The relationships among students' self-efficacy, 2 × 2 achievement goals (mastery-approach [MAp], mastery-avoidance [MAv], performance-approach [PAp], and performance-avoidance goals), and achievement performance remain largely unanswered. We tested a model of the mediating role of self-efficacy on the relationship between 2 × 2 achievement goals and cardiorespiratory fitness. Method: A sample of 276 middle school students (115 boys and 161 girls; 91 sixth graders, 87 seventh graders, and 98 eighth graders), aged 12 to 15 years (M [subscript age] = 13.34, "SD" = 0.96), responded to the Achievement Goals Questionnaire (Conroy, Elliot, & Hofer, 2003) and Self-Efficacy Questionnaire (Gao, Newton, & Carson, 2008) referenced to the fitness test. Their cardiorespiratory fitness was assessed via the Progressive Aerobic Cardiorespiratory Endurance Run (PACER) 3 days later. Results: Structural equation modeling demonstrated an acceptable model fit to the data, ?[superscript 2] (2, "N" = 105) = 1.66. Self-efficacy had a statistically significant direct effect on the PACER after controlling for the effects of the achievement goals (Upsilon[subscript "self-efficacy-PACER"] = 0.21). MAv and PAp also had direct effects on PACER performance (Upsilon[subscript "MAv-PACER"] = -0.24, and Upsilon[subscript "PAp-PACER"] = 0.24, respectively). MAp failed to exert direct effect on the PACER. However, the indirect effect of MAp on the PACER via self-efficacy was small although it was statistically significant. Additionally, the indirect effects of MAv and PAp on PACER through self-efficacy were not significant. Conclusions: Students' self-efficacy fully mediated the effect of MAp on fitness performance, as well as partially mediated the effects of MAv and PAp on cardiorespiratory fitness performance. Study implications are provided for educators and practitioners.