The evaluation of mobile and/or ubiquitous interactive systems via user testing seems a priori more relevant in the field than in a usability laboratory. However, the results of the literature are contradictory. In this article, we aim at explaining the reasons why, and we propose a methodology that could minimize biases. The experiments described in the literature and our own experiments lead us to define the interactive environment concept and three possible experimental approaches: laboratory, field and reality testing. Then, we propose a methodology and a technique -the Trojan horse- adapted to the evaluation in reality testing. At last, we illustrate the theoretical approach by three experiments and give experience feedbacks on them. We conclude on the limits of our approach.