Pupils' interest has been one of the major concerns in science education research because it can be seen as a gateway to more personalised forms of interest and motivation. However, methods to investigate situational interest in science teaching and learning are not broadly examined. This study compares the pupils' observed situational interest and their expressed situational interest. One class of Finnish fourth-graders (N?=?22, age 9-10 years) participated in a heat transfer lesson. The lesson encompassed an interactive demonstration with a thermal camera, teacher-led discussions and the conduct and presentation of a collaborative inquiry task. Pupils expressed their interest levels (scale: 1?=?very boring, 5?=?very interesting) by using an electronic response system called a "clicker". The measurement took place 15 times during the lesson, with 1 measurement being just a rehearsal. The lesson was video recorded, and visible aspects of interest at the measurement time points were analysed. Reported and observational data were compared. In most cases, the observations did not yield data compatible with the pupils' own evaluations, indicating that most pupils' expressed interest is not easily interpreted through observation of their facial expressions and behaviour. In general, the interest of the group as a whole seems to diminish during the lesson. We argue that in order to maintain and increase pupils' interest, their evaluations should be taken into account in lesson planning. Video-based research might also be further enriched and validated by employing the participants' own expressions. The clicker is a suitable means of collecting primary pupils' experiences concerning their interest levels.