[en] Construction of Indicators plays an important role in the contemporaneous society. It could be observed their ordinary use by all the activities segments. In scientific area it is not different; such practice has already been consolidated; this statement could be corroborated by the innumerous papers published about this matter in the main databases; however, there is a prevalence of quantitative studies, which obtain data from the international databases, analyzing journal articles. Concerning this finding, the main objective of this study was to elaborate scientific indicators from a local institutional data base, using as a corpus of the study the dissertations and theses, produced by a graduate program in the period of 1977 to 2009.Three types of two-dimensional indicators were obtained by using bibliometric techniques: numerical, thematic and productivity. For dissertations and theses thematic characterization it was used the Subject categories and scope descriptions and the International Nuclear Information System Thesaurus (INIS). Through the statistical technique of clustering analysis, it was possible to classify them in five main groups, showing former performance and growth future trends of each group; for data extraction about theses productivity, regarding articles published; the Curriculo Lattes, and the local institutional database were used. Co-word analysis technique was used to establish a more precise correlation, between articles and theses, and for this purpose it was used the keywords from a controlled vocabulary. In order to validate the results, it was performed a survey, with the theses authors. Results obtained indicated that 55.9% of the articles published, between 5 years, before and after the thesis presentation, are strongly correlated to it. It was also observed that, correlated articles have been published, in an average of, 1, 63 year before the thesis presentation. Concerning databases, it was shown that they are valuable tools and could be used with multifunctional purposes, with a great potential to be used in bibliometric studies. It was also observed that the indicators reliability is strongly associated to the quality of data obtained from the databases. (author)