Shed deciduous teeth were obtained from a sample of 996 New Zealand children who were participants in a longitudinal study. For each child, estimates of dentine lead were obtained from two samples of dentine. The information was used to estimate the magnitude of random errors of measurement in dentine lead values and it was estimated that between 15 and 20% of the variance in dentine lead values was attributable to random errors of measurement. In addition, dentine lead levels were influenced by small systematic errors of measurement arising from the age at which the tooth was collected and the type of tooth analysed. These sampling factors, however, accounted for less that 3% of the variance in dentine lead values. The implications of these findings for the interpretation of correlations between dentine lead levels and other variables are discussed.