In order to test the hypothesis that intelligence and suggestibility may correlate with each other differentially at different levels of IQ, research has tended to dichotomize samples into low and high IQ groups. However, the practice of artificially dichotomizing a naturally continuous variable (such as IQ) has been sternly criticized by several statistical experts. Thus, in this investigation, the possibility that the association between intelligence and suggestibility may interact with IQ level was tested directly with a quadratic transformation of the IQ scores in a sample of children (n = 158). The results demonstrated that a quadratic nonlinear association existed between intelligence and suggestibility, such that the linear association between IQ and suggestibility ceased to exist at an IQ level of approximately 105. The results are discussed in light of the advantages of not dichotomizing naturally continuously scored variables. Further, in contrast to several other investigators, we argue that intelligence and suggestibility do correlate with each other within a certain level of the normal IQ range. Copyright © 2006 John Wiley & Sons, Ltd.