Sugar overconsumption is a major health threat. For people to make healthy food choices they need to possess some knowledge about sugar; for instance, how much sugar is contained in a food item, or whether a food item contains more or less sugar than another item. Here we ask (1) how accurate is people's metric knowledge (e.g., mean, range) and mapping knowledge (i.e., relative ordering) of the sugar content of food items, and (2) can this knowledge be improved with a simple seeding intervention, in which the actual sugar content is provided for a few representative items? Participants (online experiment, N = 160) estimated the sugar content of various food items (in grams), then received feedback about the actual content for a few representative items (with or without additionally seeing the equivalent number of sugar cubes); a control group received no feedback. Finally, they estimated again the sugar content of (old and new) items. Our experiment revealed participants’ lack of metric knowledge (they overestimated mean and range of sugar content of food items) but acceptable mapping knowledge (i.e., relative ordering). Seeding improved metric knowledge for seeded and unseeded (i.e., transfer) items, and it improved mapping knowledge for seeded items. The additional visualization did not amplify the effects. A simple intervention can thus improve estimates of sugar content. Our research extends prior work on seeding interventions to the novel domain of sugar content.