Mathematics learning is shaped not only by students’ cognitive abilities but also by how they regulate their learning, perceive risk, and interpret failure within the learning process. This study developed a predictive model to examine how academic risk-taking and irrational beliefs predict metacognitive awareness among lower-secondary school students in mathematics. A quantitative correlational design was employed with 979 lower-secondary school students (ages 11–14) in Türkiye. Data were collected using three validated instruments: the Mathematics-Focused Academic Risk-Taking Scale (MFARTS), the Irrational Mathematics Beliefs Scale (IMBS), and the Metacognitive Awareness Inventory for Children (MAIC, Form B). Descriptive statistics, t-tests, ANOVA, and simple linear regression analyses were conducted. Results revealed moderate-to-high levels of metacognitive awareness overall. Academic risk-taking positively and significantly predicted metacognitive awareness, whereas irrational mathematics beliefs negatively predicted it. Fifth-grade students and those with higher mathematics achievement exhibited significantly greater metacognitive awareness and risk-taking tendencies. Irrational beliefs generally decreased with higher achievement, though certain sub-dimensions showed varied patterns across grade levels. No significant gender differences were found in metacognitive awareness or overall risk-taking. Overall, the results demonstrate that metacognitive awareness in mathematics is strengthened by students’ willingness to engage in take academic risks but weakened by rigid, irrational beliefs about learning and failure. The findings contribute to the theoretical integration of cognitive-affective models in mathematics education and provide empirical support for pedagogical approaches that promote self-regulated learning through risk-friendly classroom climates and the reframing of maladaptive beliefs.