The purpose of this study was to assess children's and adults' quantitative understanding of fractions in a new and potentially definitive way, and to determine whether this understanding was related to fraction computational skill. Thirty-seven seventh graders and 32 adults (college students) participated in the study. A computation task was used to classify subjects as "good" or "poor" at fraction computation. An estimation task was used to classify the same subjects as "good" or "poor" at understanding fraction size. If such understanding is important to learning or remembering fraction computation rules, subjects would be expected to be "good" or "poor" at both. The study indicated that a quantitative notion of fractions was only one of the variables determining fraction computational skill. (RH)