Background. Newborn screening, which identifies inherited disorders and sometimes carrier status, will increasingly involve health professionals in the provision of appropriate information and support to children and their families. The ability to understand carrier results relies on an understanding of probabilistic terms. However, little is known about how best to convey probabilistic medical information to children. Research with adult populations suggests information format significantly affects comprehension. This study aimed to explore which presentation format is most effective in conveying probabilistic information to children. Methods. A probabilistic task based on the cup game was used to measure which of five different formats was associated with greatest understanding in children aged 711 years old (n = 106). Formats used were verbal labels (e.g. rarely, sometimes), percentages, proportion-word (e.g. 1 in X), proportion-notation (e.g. 1:X) and pie charts. There was also an additional mixed format condition. In each trial a picture was presented of three cups, each with a different probability depicted beneath it, and the child was asked to select which cup was most likely to contain the ball. Three trials were presented per format. Children also rated how certain they were that they had answered correctly. Results. There was a significant relationship between format and comprehension scores. Post hoc tests showed children performed significantly better when probability was presented as a pie chart, in comparison to percentages, proportion notation, proportion-word and mixed format trials. Furthermore, most children (84%) got all trials correct for this format and children were significantly more certain that their response was correct in the pie chart trials compared to all the other formats (p <0.001). Significant positive correlations were found between self-ratings of certainty and comprehension of verbal labels, percentages and pie charts. Older age was also associated with better performance on all formats except percentages. Overall comprehension was calculated by summing the scores for the individual trials and this was independently associated with older age and higher IQ. Conclusion. The results suggest that 711 year olds can understand probability information, but that the format used will significantly affect the accuracy and confidence with which children in this age group make judgements about the likelihood of an event. Of the formats studied, pie charts appear to be the optimal method of presenting probabilistic information to children in this age group. Health professionals and designers of health messages should be cognisant of this when communicating medical information to children aged 711 years old. © 2009 Ulph et al; licensee BioMed Central Ltd.