Peer assessment is an educational activity whereby students judge the performance of their peers. It can take different forms depending on the characteristics of its implementation, the learners, and the learning context. Peer assessment has two main functions: it can be used as an assessment tool (i.e. to judge performance), and as a learning tool (i.e. to develop domain knowledge and assessment skills). While most of the research traditionally focused on peer assessment as an assessment tool, a large body of research addressing it as a learning activity has been growing for the last 20 years. Evidence from studies on both lines of research indicates that (a) the accuracy of peer assessment scores can be influenced by several characteristics of the implementation; (b) students need instructional support in order to be better assessors; (c) peer assessment can lead to improved performance for the assessor and, to a lesser extent, the assessee; (d) domain knowledge is a prerequisite for the development of peer assessment skills; (e) students’ perspective regarding peer assessment is usually negative before the implementation and does not always become more positive afterwards; and (f) interpersonal variables (e.g. trust in peers as assessors) are adversely influenced when the main purpose of peer assessment is to produce a score. Accordingly, peer assessment should be carefully implemented taking the diversity of the practices and purposes of assessment into account.