Generating quality and accurate internal feedback in self-assessment is challenging for higher education students. Assessing peer works has been linked to enhancing concurrent and follow-up self-assessments in prior research and theoretical models. However, empirical evidence concerning how peer work quality influences subsequent behaviors in online learning environments is scarce and contradictory. Hence, this study aimed to investigate the effects of the quality of the assessed peer works on writing performance and the amount, content, and accuracy of internal feedback via an online experiment. The participants were 164 university students who completed abstract writing and assessment tasks in an online setting. Specifically, they wrote an abstract for a study; provided feedback on two low-quality peer abstracts (PF− −), one high-quality and one low-quality (PF+ −) abstract, or two high-quality abstracts (PF+ +); self-scored and generated internal feedback for their abstracts; and revised their abstracts. The groups’ initial and revised abstracts, self-scorings, and internal feedback were analyzed. The findings revealed that the PF+ + and PF+ − groups had significantly higher revised writing performance, more feedback comments expressing problems in their writings and proposing specific suggestions for improvement, and more accuracy in self-scoring and problem detection than the PF− − group. Indicating the impact of the peer work quality, this study suggests using at least one high-quality work for (online) peer assessment to enhance writing performance and self-assessment accuracy.