Students in a large introductory nutrition course were divided into two groups, one that received training on estimating food quantities and one that received no instruction. Training consisted of 10-minute sessions in which 76 subjects (in groups of 10) passed around and viewed food models labeled with their portion sizes while one of the investigators verbally indicated the quantities. Immediately following instruction, the trained subjects individually estimated portion sizes of one of four displays of six foods: two solids (meat loaf and fish), two liquids (milk and soup), and two amorphous items (spaghetti and apple sauce). The second group of subjects, who had received no training, also estimated the same portion sizes of the same foods in the same displays. The four displays were set up in separate rooms, with the same foods and portion sizes in each, but different container size combinations were used (smaller vs. larger containers). Both training and type of food had a significant effect; training improved estimations, and, in general, solids were better estimated than liquids, which were better estimated than amorphous items. Although results were mixed, women tended to estimate more accurately than men. There was some tendency for estimates to be more accurate when made for foods in small containers. The authors conclude that a training session enhances the accuracy of individuals estimating food quantity and that accuracy of estimates differs for different food types.