Disagreement is essential for knowledge growth in science. However, disagreement in peer review is usually regarded as a sign of unreliability in existing studies. The predictive role of disagreement for potentially impact ful discoveries was rarely explored. Reviewer expertise has been proved to have effects on review strictness and outcomes, but reviewers' added-value on paper quality remains unclear. In this paper, we examined the predictive effects of disagreement and confidence of reviewers on citations of reviewed papers. Using a dataset of 489 papers submitted ICRL 2017, we found that predictive effects of disagreement and confidence on citations. Among accepted papers, the papers with higher review ratings tend to receive more citations and the disagreement cannot predict citations. However, among rejected papers with more than ten citations, the ratings they received cannot predict their citations, but the disagreement level of their reviewers does. Similar opposite findings were also seen regarding confidence. Accepted papers reviewed by confident reviewer tend to have more citations but rejected papers might not benefit from confident reviewers.