Background: Previous research has indicated that there may be only a modest degree of agreement between different reporters of a child's behaviour (mental health). This raises the possibility that some descriptions of the child's behaviour may reflect the personal characteristics of the respondent. We examine two potential sources of bias that may influence reports of a child's behaviour/mental health. The first is the mental or emotional impairment of the respondent; the second concerns gender-related expectations of children. Methods: Mothers (and their children after the birth) were assessed at first clinic visit, 3-5 days after the birth, then 6 months, 5 years and 14 years after the birth. Some 70% of respondents giving birth remained in the study at the 14-year follow-up, leaving some 5277 cases for this analysis. At the 14-year follow-up, child behaviour (mental health) was assessed using the Child Behaviour Check List and the Youth Self Report. Maternal mental health was determined using the anxiety and depression subscales of the Delusions-Symptoms-States Inventory. Results: Mothers who were not emotionally impaired reported fewer child behaviour problems than did the children themselves. As the mother's current emotional impairment increased, so her reports of the child's behaviour problems increased, when compared with the child's own reports. Further, mothers attributed more internalising symptoms to female respondents, and more externalising symptoms to male respondents, than did the child respondents themselves. Conclusions: Mothers differ systematically from their children when they are reporting their child's behaviour (mental health). The more emotionally impaired the mother, the greater the degree to which she imputes the child to have behaviour problems. Further, female children are attributed to have more internalising behaviours and male children externalising behaviours.