The objective of this study was to reevaluate the optimal dIle:dLys ratio during the withdrawal phase (d 42-56). Ross 708 x Ross YP male chicks were provided common starter, grower and finisher diets (70, 69, and 69% dIle:dLys, respectively) formulated using breeder recommendations, as well as optimal dIle:dLys ratios established from our previous research. On d 42, pen weights were equalized by block (12 reps/trt), and each pen contained 25 birds (0.24 m²/bird); average BW was 3.206 kg. Experimental diets were created from a common deficient diet (54% dIle:dLys; 130% dLeu:dLys; 82% dVal:dLys). Crystalline Ile was added to the deficient diet to create the summit diet (84% dIle:dLys); intermediate levels of dIle:dLys (59-79% dIle:dLys) were obtained by blending these diets. A practical control diet (69% dIle:dLys; 146% dLeu:dLys; 78% dVal:dLys) was separately formulated and batched to verify blending techniques. This study was analyzed as a randomized complete block design using Proc GLM; significance was set at P≤0.05; LS means and calculated dIle:dLys ratios were utilized for regression analyses. Birds were individually weighed on d 55 to measure CV of BW and BW, d 42-55 BW gain, feed intake, and FCR. On d 56, 5 birds/pen of average weight (±100g) were processed, and yields were calculated. Metrics not significant included CV of BW, d 55 BW, and d 42-55 feed intake (P>0.05). Numerical and significant differences were observed for d 42-55 BWG and FCR (P=0.0517 and P<0.0001, respectively); FCR improved by 13 points as dIle:dLys increased from 54 to 84% (2.046 vs 1.921, respectively). Carcass and drumstick yields were significant, whereas feeding 54% dIle:dLys increased yields (P<0.05). Preplanned contrasts between the control diets demonstrated no differences for all metrics (P>0.05), verifying blending. Optimal dIle:dLys ratios were estimated using quadratic regression (QR; 95% of vertex), as well as linear and quadratic broken line models (LBL; QBL). The QR model estimated the ratio for BW at 72% (P=0.0028, R²=0.95) and BWG at 71% (P=0.0264, R²=0.84). For LBL, the ratio was estimated to be 65% for both BW (P=0.0002, R²=0.99) and BWG (P=0.0075; R²=0.91); QBL estimated the ratio at 70% for both BW (P=0.0002; R²=0.99) and BWG (P=0.0123; R²=0.89). For FCR, the QR model estimated the ratio to be 77% (P=0.0125; R²=0.89), while LBL and QBL estimated it at 70% (P=0.0171; R²=0.87) and 78% (P=0.0118; R²=0.90), respectively.