Metapragmatic conscience enables a person to discriminate and relate linguistic and non-linguistic stimuli of an utterance (Gombert, 1992). It is essential in effective communication and its development occurs late because it emerges in school age. Research that has looked at metapragmatic conscience has included laboratory tests with a few participants, but there is no research that has employed greater populations to describe it. Given these data and within the framework provided by a project that seeks to investigate both late orality and the development of reading and writing, this paper informs about the design of some multimedia software, called Metapragmatic Conscience (MPC) that was applied to 103 boys and girls of, on average, eight years of age. Its design included three typical tests on metapragmatic conscience: Ambiguous Referent, Ambiguous Instructions (Flavell, 1993) and Observed Dialog (Gombert, 1992). Statistical analysis of the results helps conclude that the instrument has a high reliability (0.88 according to Cronbach's alpha). Findings indicate that Ambiguous Referent was the most difficult, followed by Ambiguous Instructions and Observed Dialog, owing to the difference in the context/linguistic message relation. Epi- and metapragmatic responses occur together and only does their frequency vary; this means that the passage from some to the others occurs in recursive phases, rather than in determined phases. Besides, since at eight years of age metapragmatic responses are scarce, their ontogenetic development could be described by looking at an increase in them among higher elementary and secondary school students.