A large literature uses parental evaluations of child health status to provide evidence on the socioeconomic determinants of health. If how parents perceive health questions differs by income or education level, then estimates of the socioeconomic gradient are likely to be biased and potentially misleading. In this paper we examine this issue. We directly compare child mental health evaluations from parents, teachers, children and psychiatrists for mental health problems, test whether these differences are systematically related to observable child and parent characteristics, and examine the implications of the different reports for the estimated income gradient. We find that respondents frequently evaluate children differently and while the sign of the income gradient is in the same direction across respondents, systematic differences in evaluations mean that the estimated magnitude and significance of the health-income gradient is highly dependent upon the choice of respondent and the measure of child health.