Contribution: Students do not naturally give unbiased, accurate peer assessments (PAs). Hence, giving teamwork guidelines to students can improve their cooperation, understanding of one another, PA attitude, and PA accuracy. Background: During collaborative learning, PA can improve students' autonomy, evaluation, and communication, which often enhances their task performance and learning outcomes. However, student bias, low effort, or imprecision can bias PA. Hypothesis: This study hypothesized that teamwork guidelines can improve students' collaboration and hence their: 1) PA accuracy and 2) attitudes toward PA. Design/Method: This comparison study examined the 447 PAs and 143 survey responses of 143 university students studying computer science in 49 teams. 58 students received teamwork guidelines and 85 students did not. PAs and teacher assessments (TAs) were analyzed with a multilevel, structural equation model, and survey responses were analyzed with a multivariate outcome, multilevel, and ordered logit regression. Findings: PAs often exceeded TAs of the same student (positive bias), especially for low-performing students. Compared to other students, students receiving teamwork guidelines gave PAs that resembled TAs more but had a similar positive bias. The relative and absolute differences in peer and TAs (PA - TA, |PA - TA|) mostly differed across teams rather than students. Also, students receiving teamwork guidelines viewed PA as more supportive of their learning, compared to other students without the teamwork guidelines.