Breeding values of sires resulting from selection either for reduced birth weight and increased yearling weight (YB, n = 8) or for increased yearling weight alone (YW, n = 9) were compared with each other and with sires representative of the population before selection began (BS, n = 12) using progeny testing. Reference sires (n = 6) connected these Line 1 sires with the Hereford international genetic evaluation. Thirty-five sires produced 525 progeny that were evaluated through weaning. After weaning, 225 steer progeny were individually fed, slaughtered, and carcass data collected. Data were analyzed using restricted maximum likelihood procedures for multiple traits to estimate breeding values for traits measured on the top-cross progeny while simultaneously accounting for selection of the sires. Results of the progeny test substantiate within-line results for traits upon which sires were selected. Breeding values for gestation length were greater in YB sires than in YW sires and were unchanged relative to BS sires. Breeding values for growth rate and feed intake for the YB and YW sires were greater than for BS sires. Predicted breeding values for indicators of fat deposition tended to be greater in YB sires and less in YW sires relative to BS sires, although YB and YW sires had similar breeding values for marbling score. Selection based on easily and routinely measured growth traits, although achieving the intended direct responses, may not favorably affect all components of production efficiency. Further, divergence of selection lines may not be easily anticipated from preexisting parameter estimates, particularly when selection is based on more than one trait.