Retained heterosis for different measures of gain efficiency was estimated in F3 generation castrate male progeny in three composite populations finished on two levels of dietary energy density (2.82 and 3.07 Mcal of ME and 11.50% CP) and serially slaughtered at four end points at intervals of 20 to 22 d. Breed effects were evaluated in nine parental breeds (Red Poll [R], Hereford [H], Angus [A], Limousin [L], Braunvieh [B], Pinzgauer [P], Gelbvieh [G], Simmental [S], and Charolais [C]) that contributed to the three composite populations (MARC I = 1/4 B, 1/4 C, 1/4 L, 1/8 H, 1/8 A; MARC II = 1/4 G, 1/4 S, 1/4 H, 1/4 A; and MARC III = 1/4 R, 1/4 P, 1/4 H, and 1/4 A). Gain efficiency was evaluated in time constant (0 to 207 d), gain constant (310 to 540 kg), carcass weight constant (333 kg), and retail product weight constant (225 and 210 kg) end points and to different marbling score and longissimus muscle fat end points. Expressions of gain efficiency included live weight gain/megacalories of ME and retail product weight/megacalories of ME. Significant differences were observed among breeds in all measures of gain efficiency. Breeds that had the smallest weight to maintain tended to be more efficient in live weight gain in the time constant period, whereas breeds with the highest rate of gain tended to be more efficient in the gain constant period. To marbling score or longissimus muscle fat end points, breeds with the lowest marbling scores and smallest percentage fat in the longissimus muscle on an age constant basis (e.g., Limousin and Gelbvieh) tended to be less efficient, whereas breeds with the highest marbling score and highest percentage of fat in the longissimus muscle on an age constant basis (e.g., Hereford and Angus) tended to be more efficient. Breeds with the highest percentage of retail product (Limousin and Gelbvieh) were more efficient to retail product weight end points, or when retail product weight was the measure of output. Steers fed the higher energy density diet were more efficient in live weight gain to time constant, live weight gain constant, marbling score constant, and longissimus fat constant end points and were more efficient in the production of retail product weight to retail product weight end points because of the shorter period of maintenance. Retained heterosis was not consistent among composites in different measures of gain efficiency. Higher initial weights resulting in greater requirements for maintenance had a negative effect in composites for most measures of gain efficiency.