Prosocial behavior, that is the voluntary behavior intended to benefit another, has been widely studied in children since 1970s. Though many studies focused on the age trends and gender differences in children’s prosocial behavior, no consistent conclusions have been drawn. It is suggested that developmental changes, as well as gender differences, in prosocial behavior are complex and are influenced by methodological factors. Informant discrepancies in the assessment are one of such factors but are less studied empirically. Here, the objectives of the study were to investigate agreement between mothers’ and teachers’ rating on children’s prosocial behavior, and further compared the predictive effect of different informant’s assessment on children’s actual prosocial response observed in experimental context to give implication for the cross-informant’s discrepancies. Mothers (n=86) and teachers (n=8) assessed each child (n = 86; 3-to-5 years old) using the empathic and prosocial response subscale from the conscience measure developed by Kochanska et al. (1994). Meanwhile, children’s prosocial responses to strange adult’s distress in experimental context were videotaped and coded. Mother-teacher agreement was examined through two indices; mean scores and correlation. Hierarchical regression analyses were conducted to investigate the predictive effect of different informant’s reports on children’s actual prosocial response observed in experimental context. Results showed that mean scores rated by mothers were significantly higher than those by teachers, and mother-teacher correlations were moderate for boys (0.34) but absent for girls (0.01). Further examination showed that only teacher’s rating predicted significantly boys’ prosocial response in experimental context and accounted for 11% of the variances in boys’ prosocial response scores, instead both teacher’s and mother’s rating predicted girls’ prosocial response in experimental context and accounted for 8% and 7% of the variances in girls’ prosocial response scores, respectively. Results implied that cross-informant’s discrepancies existed in assessing children’s prosocial behavior, and such discrepancies may reflect that mother and teacher emphasized different aspects of children’s prosocial behavior. On one hand, though teacher and mother paid attention to similar aspects of boy’s prosocial behavior, only teacher’s reports reflected boy’s prosocial response towards stranger