Although attention has recently been focused on understanding and preventing children's injuries, much controversy exists over the best data collection methods for examining child injuries. This study examined three methodological issues relevant to childhood injury data collection including the length of time children and parents can be expected to meaningfully recall injuries, whether the parent or child is a preferred informant about the injury, and the potential use of near injuries as a proxy measure for actual child injuries. Both children and their mothers were individually interviewed every 2 weeks for a total of 6 months about both injuries and near injuries. In addition, at the end of the 6-month period, they were asked to recall all injuries that occurred during those 6 months. Overall, children reported more injuries than mothers. Children recalled far fewer and mothers recalled slightly fewer events than had been reported in the biweekly interviews. There were fewer near injury than actual injury events reported, although this varied across categories, with some categories (e.g., car passenger injuries) having more near than actual injuries, and other categories (e.g., cuts, bumps, and bruises) having many more actual than near injuries reported. Limitations of the project are discussed and implications for future research advanced.