Background. Good teachers have been studied ever since Plato described how Socrates taught by asking questions of his audience. Recent findings shed light on two characteristics of good teachers: their personality and their ability. However, more attention has been paid to teachers' practices and opinions than to students' views. Aims. The study reported here attempted to deepen our understanding of what students think about good teachers. Sample. Students of four age groups (7, 10, 13, and 16 years of age) and teachers from primary and secondary schools were asked to write an essay on the good teacher. Methods. The correspondence between conceptual items in the essays was investigated by determining the extent to which they were used in the same essays to describe good teachers. Results. Correspondence analysis revealed two dimensions. The first dimension reflected the preference of students and teachers for describing the good teacher in terms of either personality or ability characteristics. The second dimension was interpreted as an orientation in the essays towards either attachment to, detachment from or commitment to school and teachers. Students and teachers were compared to establish the amount of (dis)agreement about what makes a good teacher. Primary school students described good teachers primarily as competent instructors, focusing on transfer of knowledge and skills, whereas secondary school students emphasised relational aspects of good teachers. Teachers, however, considered good teachers in the first place a matter of establishing personal relationships with their students. Consequently, primary school students and teachers disagreed about the characteristics of good teachers. In secondary education, disagreements between teachers and students were relatively small. Conclusions. The research method of collecting free essays and utilising correspondence analysis to represent conceptual items and groups of participants seems promising as long as a theoretical framework is available to interpret the resulting representation of similarities between items and groups of participants.