This paper explores the hypothesis that there is an association between risk of childhood kidney cancer and paternal employment in occupations that have potential for exposure to herbicides and/or pesticides. In contrast to a previous study using paternal occupations at the time of the child's death, no significant association was found between such potential paternal exposures (using paternal occupations at the time of the child's birth) and the risk of cancer of the kidney in childhood (OR 0.88, 95% CI 0.20-3.84). In addition, this paper quantifies the extent to which paternal occupational information on death certificates can be used as a proxy for paternal occupation at birth and how such misclassification could affect risk estimates. An example is given showing how a recently reported association between kidney cancer and paternal agricultural employment may have been overestimated as a result of the use of death certificate information.