Metacognition refers to the capacity to reflect upon our own cognitive processes. Its contribution to reading development, when children start building their orthographic lexicon, still remains unknown. Here, we evaluate the metacognitive efficiency of children aged between 6 and 7 years old (N = 60) in 5 experimental tasks; four linguistic tasks assessing orthographic lexical processing and a non-linguistic task unrelated to reading skills. First, we investigated how metacognition on the experimental tasks related to standardised on-paper reading performance, hence participants' general reading level. Second, we assessed whether these developing readers recruited common metacognitive mechanisms across the different experimental tasks. Third, we explored whether metacognition in this early stage was related to the longitudinal improvement in performance on a linguistic vs a non-linguistic task. No association was found between students' metacognition in the reading-related tasks and performance on the standardised reading tests, notwithstanding first-order performance correlated across these tasks. Remarkably, some negative correlations were noted between students' metacognitive ability in one task and task performance in another task. Moreover, we found some evidence consistent with shared metacognitive mechanisms for monitoring performance across tasks. Finally, metacognitive efficiency significantly predicted children's performance improvement across domains 10 months later. These results suggest that the development of metacognitive processing may be dissociated to some extent from reading-related linguistic abilities during the early stages of formal education. Nevertheless, it may play a fundamental role in guiding students' learning across domains. These data highlight the importance of creating educational programs fostering students' metacognition as a long-term learning tool.