The aim of the study was to investigate how Swedish teachers and student teachers of Swedish vary in their assessment of a pupil’s text, in judging both the text as a whole and its parts. In the study the informants were asked to assess a pupil’s text (a text presenting an argument) from the national examination in Swedish 1 with the aid of an assessment matrix and then complete a questionnaire assessing the text as a whole and in its parts. The responses to the questionnaire were analysed statistically with a focus on how the assessments varied. A sociocultural and situated perspective was applied in the study, along with psychometry and the theory of discourses of writing, to give more perspectives on the results. The study found that the assessments varied between grades C and A for the text as a whole and between E and A for the aspects of argumentation, content and critical reading, contextual signals and reference connectors, introduction, conclusion, situational adaptation and self-sustaining. The study also found that the student teachers were generally stricter in their assessment. The greatest variations in the student teachers’ assessment were found to concern aspects of the communication discourse, while in the teachers’ assessment it was aspects of the construction discourse that revealed the greatest variation. The greatest difference between the two groups concerned aspects of the correction discourse.