Taking into account affects in the design and evaluation of technologies has become an important topic during the last few years. Nevertheless, this approach still raises numerous theoretical and methodological questions. This article aims at identifying some of the foundations of such an approach. The first part examines the contribution of affects in the evaluation of technologies through an analysis of the main models in the field: usability model, technology acceptance model and user experience. It is shown that user experience seems the most relevant approach to giving an account of the complexity of affective reactions in human-computer interactions. It is also shown that this approach has some limits. The second part of the article is focused on the relation between pragmatic qualities of the system (usability) and users? affective reactions: for example, whether they are independent, complementary and so on. Finally, methodological issues are examined. The choice of affects measurement tools is a critical point. In fact, knowledge about users? affective reactions in human-computer interactions is still limited. Moreover, the capture of dynamic emotions appears to be important, as does having access to emotion antecedents or factors. In order to contribute to these issues, the benefits and limits of different kind of affects measures are discussed, as non verbal measures, combined physiologic measures and a questionnaire, and interviews. The final part presents perspectives on the study of affects in human-computer interactions.