<p>The development of metacognition is a critical focus within the domain of metacognitive research. Prior studies have established a significant association between metacognition and mathematics achievement. However, the dynamic interplay between these constructs remains underexplored. To elucidate this relationship, the current study employed a longitudinal survey design, tracking 334 junior high school students in Jiangxi Province, China, over a 2.5-year period with biannual assessments. This study aimed to examine the developmental trajectory of metacognition among junior high school students and its impact on mathematics achievement. Results indicated that the developmental pattern of metacognition among junior high school students was characterized by an initial increase during the first year, followed by declines in the second and third years. Latent class analysis identified three distinct developmental trajectories: the <i>High-slow-rising group</i> (26.67%), the <i>Middle-slow-declining group</i> (54.85%), and the <i>Low-</i>. <i>significant-declining group</i> (18.48%). Furthermore, the intercept and slope of metacognition significantly predicted the intercept of mathematics achievement, but not its slope. Regarding the developmental trajectories of mathematics achievement, metacognition and its dimensions exhibited significant negative growth in the <i>Middle-significant-declining group</i> and the <i>Low-</i>. <i>significant-rising group</i>, whereas growth in the other two groups was not significant. These findings highlight the heterogeneous developmental patterns of metacognition among junior high school students and reveal a complex dynamic relationship between metacognition and mathematics achievement. The results underscore the importance of considering individual differences in metacognitive development when examining its impact on academic achievement.</p>