This article deals with evaluating an interactive learning environment (ILE) during the iterative-design process. Various aspects of the system must be assessed and a number of evaluation methods are available. In designing the ILE Ambre-add, several techniques were combined to test and refine the system. In particular, we point out the merits of combining measures of the ILE's impact on learning with analyses of the learner's activity. In this article, the approach used to evaluate Ambre-add is described and analyzed, and its advantages and disadvantages are discussed. Then, a general procedure for evaluating ILEs during iterative design is proposed. (Contains 4 figures, 2 tables and 1 note.)