Categories for evaluation of educative software by different authors are analyzed. For STI theoretical basis are: learn by doing, construction of knowledge, mental conflict, zone of proximal development, learning by discovery, conceptual change, and learning for the understanding and the standards for interfaces. A proposal of evaluation based on the usability criteria is presented and is add new criteria to the existing categories and new categories of analysis are generated. The application dominion causes that some special characteristics are due to have in account as the design of the exercises and problems and in the system of self-evaluation and evaluation