We proposed an hypothesis that ruminants attempt to select a diet that promotes high levels of feed intake by maintaining optimal ruminal conditions. Three tests of the hypothesis considered whether the diet selection of sheep given either a choice of two high energy density (ED) feeds or a choice between a high and a low ED feed is affected by 1) sodium bicarbonate inclusion (NaHCO3; 1, 2, and 4% [wt/wt]) in the high [ED] feed); 2) dietary carbohydrate source (barley-based, B and sugar beet/ barley, S) of the high ED feed; or 3) physical form (alfalfa: pelleted, ALFP and long chop, ALFL) of the low ED feed. To conduct these, 42 lambs were used in seven 6 x 6 Latin squares, which were either foods ALFL and ALFP offered alone and paired with feed B or S and their NaHCO3 derivatives as a choice, or foods B and S offered alone and paired with their NaHCO3 derivatives as a choice. Each Latin square period lasted for 3 wk. For the choices between a NaHCO3-supplemented high ED feed and either a high or a low ED one, the inclusion of NaHCO3 increased feed intake. Its level of inclusion had a marked effect on the proportion of unsupplemented high ED feed selected (P < .05) but not on the proportion of low ED feed (alfalfa) selected. More alfalfa was selected when the high ED feed was based on barley (B) rather than sugar beet/barley (S) (P < .05). The physical form of the alfalfa affected diet selection because its selection was greater (P < .001) when ALFP (pelleted) was offered in the feed choice rather than ALFL (long-chop). The significant effects of NaHCO3 level, dietary carbohydrate, and physical form on the diet selection and feed intake of the sheep are consistent with the proposed experimental hypothesis.