The effects of roughage level (10, 20, or 30% roughage equivalent [RE]), roughage source (alfalfa vs cottonseed hulls), roughage regimen (constant RE vs 2% RE during the mid-finishing period), tallow level (1.2 vs 4.6%), and steer type (British crossbred [BRITX] vs Bos indicus crosses [BRX]) were evaluated in three experiments with a common allotment and several overlapping treatments. Steers (n = 432; initial weight = 326 +/- 26 kg) were divided into three BW blocks and allotted randomly to 72 pens and 24 treatments. Steers were fed steam-flaked, sorghum grain-based finishing diets for 124 to 166 d. Diets with 20% RE decreased gain efficiency and 30% RE diets decreased both gain (linear, P < .07) and efficiency (linear, P < .001) compared with 10% RE diets. Reducing roughage level during the mid-finishing period improved overall gain efficiency 2, 7, and 24% (P > .2, < .05, and < .001, respectively) for the 10, 20, and 30% RE diets, respectively. Steers fed cottonseed hulls consumed more feed (9.6 vs 8.8 kg/d, P < .001) but tended to gain less (1.53 vs 1.58 kg/d, P = .11) than steers fed alfalfa, were leaner, and had fewer carcasses grading Choice (62 vs 77%, P < .05). Feeding 4.6% tallow decreased DMI (P < .05) and improved gain efficiency (P < .05) compared with 1.2% tallow. The BRITX steers consumed more feed (6%, P < .001) but were somewhat less efficient (3.5%, P < .05) than BRX steers. Carcasses from BRITX steers tended to be fatter than carcasses from BRX steers and more of them graded Choice (62 vs 37%, P < .01). Commercial BRX steers did not perform as well as BRITX steers on higher-energy-density diets (4.6% tallow or variable roughage regimen). Knowledge of the genetic background of feeder cattle can be important in the selection of dietary energy density and marketing expectations.