Although peer assessment is a widely used didactical method in higher education, little is known about a) how many peer assessors are required to receive a stable assessment on a peer’s solution and b) how valid this assessment is compared to an expert assessment. To fill these gaps, we conducted a peer assessment in a large-scale lecture. Overall, 136 students participated in the peer assessment. Each student had to complete an assignment, which was then anonymously evaluated by five randomly selected peers, and three independent experts. We applied a bootstrap-based Monte Carlo simulation based on our data. The results show that a) three peer assessors are sufficient for a stable assessment, and b) the peer assessors are less critical compared to experts. We thus contribute to literature by providing insights on how many peer assessors are required when applying peer assessment, and how comparable peer assessment is with expert assessment.