Changes in performance traits in beef cattle over the last 30 yr necessitate the re-evaluation of central testing procedures to ensure that they are cost effective and appropriate for current test objectives. The objective of this study was to evaluate the potential for reducing the length of evaluation from 140 d to either 112 or 84 d. Data evaluated were postweaning feedlot performance test records (collected from 1977 to 1986) on 1,830 individually fed bulls. Bulls representing 13 breeds (n greater than 25 per breed) were evaluated in the University of Arkansas Cooperative Bull Tests at Fayetteville, Hope, and Monticello, Arkansas. Models were fit for ADG, daily feed intake (FI) and feed conversion (FCONV) from d 1 to d 140 (ADG1-140, FI1-140 and FCONV1-140, respectively) and from d 1 to d 112 (ADG1-112, F11-112 and FCONV1-112, respectively). Models fit for ADG1-140, FI1-140, and FCONV1-140 using information up to d 112 had R2 of .90, .99, .88, and .94, respectively, and using information up to d 84 had R2 of .82, .94, and .80, respectively. Spearman rank correlation coefficients (all P less than .0001) were .93 for ADG1-140 and ADG1-112, .93 for ADG1-112 and ADG1-84, .99 for FI1-140 and FI1-112, .91 for FCONV1-140 and FCONV1-112, and .90 for FCONV1-112 and FCONV1-84. These coefficients indicate that bulls ranked similarly for performance traits at 84, 112, and 140 d. If the primary objective of central station testing is to evaluate ADG during the linear phase of growth, and if testing procedures are those currently recommended by the Beef Improvement Federation, then feeding bulls beyond 112 d has no advantage because the information upon which selection decisions are made is similar at both 112 and 140 d.