Three dietary compositions and 3 methods of pellet preparation were used in different choice situations to improve our understanding of the factors perceived by broiler chickens in the physical structure of their diet. Within the same visual field, broiler chickens were offered 8 microfeeders containing different pelleted feeds that could be reached from the same position of the bird. Chickens were daily tested during 5-min tests following a 45-min period of feed withdrawal, after adaptation to individual cages and experimental feeders. Two feeds (4 microfeeders/feed in alternate positions) were weighed prior to and after the test. Behavior of the chickens was videotaped during the test to count the number of pecks given to each microfeeder. In a first experiment, 2 feeds of distinct composition were either steam or cold-pelleted. The tests of choice were repeated daily from 22 to 25 and from 35 to 38 d of age on 96 male broiler chickens. Significant preferences were measured from the first test. Similar preferences were measured when chickens received the same choice or alternating choices from one test to another. Differences of feed intake between 2 feeds during 5 min were significantly correlated to differences of number of pecks (R2 = 0.85). In most cases, preferences were established within the first minute of pecking. Steam pellets were significantly preferred to cold pellets for one feed composition but not for the other. Preferences for compositions were consistent whatever the pellet preparation and probably associated with the color of the feeds. A second experiment confirmed that 38-d-old broiler chickens seemed to prefer light colors. Short-term preferences of a chicken for a feed were rather influenced by physical characteristics of feed particles (size, color, hardness) than by the origin of raw materials. Feed pecking may change because of feed composition itself, because of its consequences on color or texture cues, or both.