It is well established that teachers perform evaluations or assessments of student answers. This chapter shows that participants orient to three different dimensions of evaluations: the <i>positive/negative</i> dimension, the <i>object</i> dimension of what is being assessed, and the dimension of the <i>value</i> according to which the object is assessed. The paper then focuses on the object dimension of teacher evaluations in a data set of dyadic teacher-student explanations in mathematics classrooms. The vast majority of teacher evaluations is either concerned with students’ ‘knowing’, or with students’ ‘doing’ – for example when the teacher treats a wrong answer as a mistake -, not with students’ ‘understanding’. The chapter shows both the sequential characteristics of these differently oriented evaluations, and their turn formats.