Parent-assigned pain scores have been used as proxy measures of pain for children, such as those with cognitive impairment (CI), who cannot self-report. However, the accuracy of parent-assigned pain ratings for children with CI has not been studied. This study evaluated the construct and criterion validity of parental pain scores of children with CI. Fifty-two children aged 4 to 19 years with CI and their parents/guardians were included in this observational study. Children were observed and assessed for pain by parents using the Faces, Legs, Activity, Cry, and Consolability (FLACC) observational tool and the 0 to 10 Numbers Scale, and simultaneously by nurses using the FLACC. Children who were cognitively able scored pain using simplified scales. Parent scores decreased after analgesic administration (6.4 +/- 2.5 vs. 3.1 +/- 2.3; p = .004), supporting their construct validity. Parents' FLACC and Numbers ratings correlated well with nurse ratings (intraclass correlation coefficient = 0.78 [confidence interval = 0.63-0.87] and intraclass correlation coefficient = 0.73 [confidence interval = 0.59-0.83], respectively). The parents' coded Numbers ratings correlated moderately with their child's ratings (rho = 0.57; p = .05) and agreed in 20% to 100% of cases (kappa = 0.388). There was better overall agreement between parents' FLACC scores and child ratings (33%-67% agreement; kappa = 0.43). The parent underestimated the child's pain with FLACC ratings in only one case (8%), but overestimated pain in three cases (25%). This study suggests that parents of children with CI provide reasonable estimates of their child's pain, particularly when using a structured pain tool. Parents may, however, tend to overestimate their child's pain during the early postoperative period.