This research presents results of an exploratory study with 14 university students with the objective of evaluating if abstracts can be taken as parameters of reading comprehension. For Kintsch (2002), the successful reader is the one that incorporates information from the text with their knowledge of the world. The issue is how to investigate reading comprehension, since there is no direct access to mental representations and operations involved in this process. Based on the hypothesis that abstracts may exemplify the overall semantic representation of a text (van Dijk, 2004), we investigated the summarization operations favored by the subjects in the production of summaries of a text. The research also aimed to contribute to the evaluation of techniques for automatic text summarization. We analyzed summaries resulted from the highlight of the main ideas of the text, and also from free written summaries of the same text. These texts were compared to a summary produced by the summarization program GistSumm (PARDO, 2002). Taking into consideration the syntactic clues for the student’s choice of maintenance and deletion of the sentences highlighted in the summaries, there was no uniformity of choices; main differences occurred in complex sentences. A qualitative analysis related to the propositions, connectives and coreference chain in the student’s free summaries allowed us to assess the degree of understanding of the source text by the subjects of this research. We believe that the results can be interpreted as indicative that abstracts could be an important resource for the assessment of reading comprehension.