This thesis investigated the measurement of emotion during short episodes of interaction between products and their users. Chapter 2 is a review of the many ways that have been used to measure emotions, organized according to the component of emotion involved: feelings, bodily changes, and facial expression. Measurement based on bodily changes and facial expression is costly and requires extensive expertise. Still, several physiological measures have been considered in the design-related literature but they often lack specificity. Even if automatic recognition systems have recently become available, applied research based on the observation of facial expression remains extremely rare. Both physiological recording and facial expression recognition could in principle have huge advantages for moment-to-moment assessment of emotion as they provide nearly continuous data without requiring the active participation of the research participants. However, their lack of reliability forces researchers to rely on multiple trials and averaging in analysis, thus precluding simple online measurement. Self-report, based on conscious feelings, is easier to apply and is the most common way to measure emotions. Self-report measurement instruments based on different models of emotion are available including measures of pleasantness and arousal and measures of discrete emotions like anger or disgust. Several of these questionnaires have been used in a design context, often to assess responses to product appearance or long-term use. Moment-to-moment self-report is also common in fields like advertisement or music research but is typically limited to dimensional models of emotion (measuring pleasantness or arousal). Chapter 3 is devoted to punctual measures of emotion in person-product interaction. It describes two studies in which participants had to complete different questionnaires right after using a product. The first study compared two questionnaires chosen for their extensive coverage of positive emotions – PrEmo and the Geneva Emotion Wheel – in a test with a coffee machine and an alarm clock. The results show both instruments to be sensitive to differences between products and document a decent level of convergence between the questionnaires. The second study extended these results to a between-subject experimental design in which each participant only used one of the products tested. It found a variant of PrEmo to be sensitive to differences between several personal navigation devices and examined the relationships between measures of different aspects of user experience (perceived usability, meaning, feelings). Chapter 4 is devoted to continuous or moment-to-moment measures of emotion in person-product interaction. It describes the particular challenges facing researchers interested in the dynamics of ongoing emotional changes during the interaction itself. It then sketches an approach developed to tackle this problem, by combining several techniques used in other fields. A key element of this approach is a technique called self-confrontation. It uses video to collect time-bound data about specific events right after the interaction while avoiding interrupting as it unfolds. Chapter 5 describes two studies conducted with the approach developed in chapter 4. The first study asked participants to report about their experience using two vases, selected to be either frustrating or surprising. The second study collected data about the pleasantness or unpleasantness of a drive using one of several personal navigation devices. The differences between the products were found to be related to specific parts of the routes the participants had to follow. The results also suggest that the peak experience (how bad the experience was at its worse or how good it was at its best) is more important in determining the overall experience than the average experience over the whole test. Chapter 6 describes the development of a device, the emotion slider, conceived to make moment-to-moment self-report more intuitive following the principles of tangible design. An experiment using pictures as affective stimuli was conducted before using the emotion slider to collect moment-to-moment data about dynamic stimuli. Following some unexpected results, a series of experiments was organized to better understand the properties of the slider. These experiments showed that the link between movement and affect is more complex than initially thought. Chapter 7 discusses reliability and its impact for applied measurement. It starts with a brief review of key concepts and of the limitations of some common measures of reliability. A numerical example shows that these measures can be misleading when improperly applied to data about transient states like product-related emotions as opposed to individual traits like personality and intelligence. Generalizability theory, a technique that can be used to deal with these issues is introduced through a re-analysis of some the data from chapter 3. Chapter 8 is devoted to the notion of measurement validity. After a review of the most salient perspectives on validity within psychometrics, the data presented in chapters 3 and 5 are re-evaluated. The chapter also contains a discussion of several conceptual issues regarding the validity of measures derived from different components of emotion.