During the last decades, assessment of written discourse comprehension has attracted the interest of a large number of specialists from various disciplinary fields, such as psycholinguistics, discourse psychology, applied linguistics, among others. This has led to a significant number of studies, most of which are essentially focused on issues regarding assessment instruments. Such issues include, among other aspects, how pertinent assessment techniques are, how important text's features are in the measuring, the number and the type of procedures required for validating instruments. However, in spite of the great variety of studies in the area, frameworks to guide the creation of assessment tools from a clear theoretical-methodological perspective are scarce. In this paper, we propose a framework for the design and construction of instruments for assessing written text comprehension, based on our theoretical conception of the phenomenon. Besides, in this proposal, we emphasize the importance of corpus studies to support the specialist's decision making during the design and construction of the instrument. Corpus studies can guide the generation and maintenance of a coherent relation among the theoretical constructs to be measured, the texts to be selected, and the assessing techniques to be used. In turn, this should result in a better assessment instrument.